30-op2scp-random-forest-with-tabmod-embeds.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "27a25cde",
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:39.538272Z",
     "iopub.status.busy": "2023-12-08T18:55:39.537682Z",
     "iopub.status.idle": "2023-12-08T18:55:45.198413Z",
     "shell.execute_reply": "2023-12-08T18:55:45.197759Z"
    },
    "papermill": {
     "duration": 5.671476,
     "end_time": "2023-12-08T18:55:45.200301",
     "exception": false,
     "start_time": "2023-12-08T18:55:39.528825",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import time\n",
    "t0start = time.time()\n",
    "from fastai.collab import *\n",
    "from fastai.tabular.all import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "637cdf15",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:45.216060Z",
     "iopub.status.busy": "2023-12-08T18:55:45.215501Z",
     "iopub.status.idle": "2023-12-08T18:55:45.218557Z",
     "shell.execute_reply": "2023-12-08T18:55:45.218029Z"
    },
    "papermill": {
     "duration": 0.012625,
     "end_time": "2023-12-08T18:55:45.220070",
     "exception": false,
     "start_time": "2023-12-08T18:55:45.207445",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "random_seed = 209"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "84caa667",
   "metadata": {
    "papermill": {
     "duration": 0.0064,
     "end_time": "2023-12-08T18:55:45.233363",
     "exception": false,
     "start_time": "2023-12-08T18:55:45.226963",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Loading and Melting Train and Test Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e5e89653",
   "metadata": {
    "papermill": {
     "duration": 0.006389,
     "end_time": "2023-12-08T18:55:45.246420",
     "exception": false,
     "start_time": "2023-12-08T18:55:45.240031",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Here I read the training and test data and melt it to yield a ```DataFrame``` with three categorical features (```cell_type```, ```sm_name```, and ```gene```) and one target (```value```)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "8ea22d0b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:45.261360Z",
     "iopub.status.busy": "2023-12-08T18:55:45.260851Z",
     "iopub.status.idle": "2023-12-08T18:55:48.807855Z",
     "shell.execute_reply": "2023-12-08T18:55:48.807036Z"
    },
    "papermill": {
     "duration": 3.556432,
     "end_time": "2023-12-08T18:55:48.809512",
     "exception": false,
     "start_time": "2023-12-08T18:55:45.253080",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2.78 s, sys: 555 ms, total: 3.34 s\n",
      "Wall time: 3.54 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "fn = '/kaggle/input/open-problems-single-cell-perturbations/de_train.parquet'\n",
    "df_de_train = pd.read_parquet(fn)\n",
    "train_df = df_de_train.melt(id_vars=['cell_type', 'sm_name'], value_vars=df_de_train.iloc[:,5:].columns, var_name='gene', value_name='value')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "8d4e3012",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:48.825486Z",
     "iopub.status.busy": "2023-12-08T18:55:48.825186Z",
     "iopub.status.idle": "2023-12-08T18:55:53.018320Z",
     "shell.execute_reply": "2023-12-08T18:55:53.017436Z"
    },
    "papermill": {
     "duration": 4.203741,
     "end_time": "2023-12-08T18:55:53.020255",
     "exception": false,
     "start_time": "2023-12-08T18:55:48.816514",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "fn = '/kaggle/input/open-problems-single-cell-perturbations/id_map.csv'\n",
    "df_id_map = pd.read_csv(fn)\n",
    "fn = '/kaggle/input/open-problems-single-cell-perturbations/sample_submission.csv'\n",
    "df = pd.read_csv(fn, index_col = 0)\n",
    "\n",
    "cols_to_add = df_de_train.iloc[:,5:].columns\n",
    "cols_to_add\n",
    "\n",
    "df_zeros = pd.DataFrame(0.0, columns=cols_to_add, index=df_id_map.index)\n",
    "df_zeros\n",
    "\n",
    "df_id_map_preds = pd.concat([df_id_map, df_zeros], axis=1)\n",
    "test_df = df_id_map_preds.melt(id_vars=['cell_type', 'sm_name'], value_vars=df_id_map_preds.iloc[:,3:].columns, var_name='gene', value_name='value')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e782883c",
   "metadata": {
    "papermill": {
     "duration": 0.006524,
     "end_time": "2023-12-08T18:55:53.033860",
     "exception": false,
     "start_time": "2023-12-08T18:55:53.027336",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Load Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5a8ce7bb",
   "metadata": {
    "papermill": {
     "duration": 0.006679,
     "end_time": "2023-12-08T18:55:53.047273",
     "exception": false,
     "start_time": "2023-12-08T18:55:53.040594",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "The function ```reduce_emb_dim``` is used to reduce the dimensionality of molecular (```dim = 26```) and gene embeddings (```dim = 1000```) down to 10 components."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "04e95435",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:53.062183Z",
     "iopub.status.busy": "2023-12-08T18:55:53.061881Z",
     "iopub.status.idle": "2023-12-08T18:55:53.159156Z",
     "shell.execute_reply": "2023-12-08T18:55:53.158275Z"
    },
    "papermill": {
     "duration": 0.107321,
     "end_time": "2023-12-08T18:55:53.161402",
     "exception": false,
     "start_time": "2023-12-08T18:55:53.054081",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.decomposition import PCA\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "def reduce_emb_dim(data, n_comp=35, random_state=42):\n",
    "    embname = data.columns[1][:-1]\n",
    "    Y = data.iloc[:,1:]\n",
    "    scaler = StandardScaler()\n",
    "    Y_std = scaler.fit_transform(Y)\n",
    "    reducer = PCA(n_components=n_comp, random_state=random_state)\n",
    "\n",
    "    Yr = reducer.fit_transform(Y_std)\n",
    "    column_names = [f'{embname}pca{n_comp}_{i}' for i in range(n_comp)]\n",
    "    reduced_data = pd.DataFrame(Yr, columns = column_names)\n",
    "    reduced_data = pd.concat([data.iloc[:, 0], reduced_data], axis=1)\n",
    "    return reduced_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d3ebc373",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:53.176994Z",
     "iopub.status.busy": "2023-12-08T18:55:53.176328Z",
     "iopub.status.idle": "2023-12-08T18:55:54.126531Z",
     "shell.execute_reply": "2023-12-08T18:55:54.125920Z"
    },
    "papermill": {
     "duration": 0.960008,
     "end_time": "2023-12-08T18:55:54.128534",
     "exception": false,
     "start_time": "2023-12-08T18:55:53.168526",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "cell_embs = pd.read_csv('/kaggle/input/op2-single-cell-perturbations-tabmodnn-embeddings/cell_embeddings_no_pca.csv', index_col = 0)\n",
    "mol_embs = pd.read_csv('/kaggle/input/op2-single-cell-perturbations-tabmodnn-embeddings/molecular_embeddings_no_pca.csv', index_col = 0)\n",
    "gene_embs = pd.read_parquet('/kaggle/input/op2-single-cell-perturbations-tabmodnn-embeddings/gene_embeddings_no_pca.parquet')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "f38a710e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:54.143540Z",
     "iopub.status.busy": "2023-12-08T18:55:54.143132Z",
     "iopub.status.idle": "2023-12-08T18:55:54.915275Z",
     "shell.execute_reply": "2023-12-08T18:55:54.914550Z"
    },
    "papermill": {
     "duration": 0.781788,
     "end_time": "2023-12-08T18:55:54.917416",
     "exception": false,
     "start_time": "2023-12-08T18:55:54.135628",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "gene_embs = reduce_emb_dim(gene_embs, n_comp=10, random_state=random_seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "397edaa6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:54.954333Z",
     "iopub.status.busy": "2023-12-08T18:55:54.953523Z",
     "iopub.status.idle": "2023-12-08T18:55:54.970027Z",
     "shell.execute_reply": "2023-12-08T18:55:54.969214Z"
    },
    "papermill": {
     "duration": 0.045066,
     "end_time": "2023-12-08T18:55:54.972008",
     "exception": false,
     "start_time": "2023-12-08T18:55:54.926942",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "mol_embs = reduce_emb_dim(mol_embs, n_comp=10, random_state=random_seed)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d06a639",
   "metadata": {
    "papermill": {
     "duration": 0.01634,
     "end_time": "2023-12-08T18:55:55.008477",
     "exception": false,
     "start_time": "2023-12-08T18:55:54.992137",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "c0a57943",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:55.023285Z",
     "iopub.status.busy": "2023-12-08T18:55:55.022966Z",
     "iopub.status.idle": "2023-12-08T18:55:55.027011Z",
     "shell.execute_reply": "2023-12-08T18:55:55.026351Z"
    },
    "papermill": {
     "duration": 0.013536,
     "end_time": "2023-12-08T18:55:55.028787",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.015251",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def splitter(df):\n",
    "    train = df.index[~df['is_valid']].tolist()\n",
    "    valid = df.index[df['is_valid']].tolist()\n",
    "    return L(train), L(valid)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e1477244",
   "metadata": {
    "papermill": {
     "duration": 0.00687,
     "end_time": "2023-12-08T18:55:55.042731",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.035861",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "The function ```get_train_valid_data``` takes the train and test data as inputs and returns a ```TabularPandas``` object using the ```valid``` scheme and ```split_func``` provided."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "027b604b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:55.057746Z",
     "iopub.status.busy": "2023-12-08T18:55:55.057219Z",
     "iopub.status.idle": "2023-12-08T18:55:55.065357Z",
     "shell.execute_reply": "2023-12-08T18:55:55.064760Z"
    },
    "papermill": {
     "duration": 0.017406,
     "end_time": "2023-12-08T18:55:55.066950",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.049544",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def get_train_valid_data(train_data, test_data, valid=None, random_state=None, split_func=splitter):\n",
    "    train_data['is_valid'] = False\n",
    "    cont, cat = cont_cat_split(train_data, 1, dep_var='value')\n",
    "    if valid in ['NK cells', 'T cells CD4+', 'T cells CD8+', 'T regulatory cells']:\n",
    "        test_compounds = test_data['sm_name'].unique().tolist()\n",
    "        valid_indices = train_data.loc[(train_data['cell_type']==valid) & train_data['sm_name'].isin(test_compounds)].index.sort_values().tolist()\n",
    "        train_data.loc[valid_indices, 'is_valid'] = True\n",
    "        splits = splitter(train_data)\n",
    "        cont, cat = cont_cat_split(train_data, 1, dep_var='value')\n",
    "        cat.remove('is_valid')\n",
    "        to = TabularPandas(train_data, procs=[Categorify, FillMissing], cont_names=cont, cat_names=cat, y_names='value', splits=splits)\n",
    "    elif type(valid)==float:\n",
    "        valid_indices = train_data.sample(frac=valid, random_state=random_state).index.sort_values().tolist()\n",
    "        train_data.loc[valid_indices, 'is_valid'] = True\n",
    "        splits = splitter(train_data)\n",
    "        cat.remove('is_valid')\n",
    "        to = TabularPandas(train_data, procs=[Categorify, FillMissing], cont_names=cont, cat_names=cat, y_names='value', splits=splits)\n",
    "    else:\n",
    "        cat.remove('is_valid')\n",
    "        to = TabularPandas(train_data, procs=[Categorify, FillMissing], cont_names=cont, cat_names=cat, y_names='value', splits=None)\n",
    "    return to"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "daf7a4e3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:55.081776Z",
     "iopub.status.busy": "2023-12-08T18:55:55.081491Z",
     "iopub.status.idle": "2023-12-08T18:55:55.085688Z",
     "shell.execute_reply": "2023-12-08T18:55:55.084914Z"
    },
    "papermill": {
     "duration": 0.013479,
     "end_time": "2023-12-08T18:55:55.087411",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.073932",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def rmse(preds, targs): return round(math.sqrt(((targs-preds)**2).mean()), 6)\n",
    "def m_rmse(m, xs, y): return rmse(m.predict(xs), y)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a7f382de",
   "metadata": {
    "papermill": {
     "duration": 0.006362,
     "end_time": "2023-12-08T18:55:55.100499",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.094137",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "The ```rf``` function trains a random forest and facilitates quick experimenting."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "e2566e5e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:55.115463Z",
     "iopub.status.busy": "2023-12-08T18:55:55.114921Z",
     "iopub.status.idle": "2023-12-08T18:55:55.214809Z",
     "shell.execute_reply": "2023-12-08T18:55:55.213932Z"
    },
    "papermill": {
     "duration": 0.109845,
     "end_time": "2023-12-08T18:55:55.216987",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.107142",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "def rf(xs, y, n_estimators=100, max_samples=0.66, max_features='sqrt', min_samples_leaf=5, **kwargs):\n",
    "    return RandomForestRegressor(n_jobs=-1, n_estimators=n_estimators, max_samples=max_samples, max_features=max_features, min_samples_leaf=min_samples_leaf, oob_score=True).fit(xs, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f7844c3b",
   "metadata": {
    "papermill": {
     "duration": 0.00635,
     "end_time": "2023-12-08T18:55:55.230289",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.223939",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Add Embeddings"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a3f814de",
   "metadata": {
    "papermill": {
     "duration": 0.006334,
     "end_time": "2023-12-08T18:55:55.243250",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.236916",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "The following lines merge the embeddings with the train and test data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "dc912c44",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:55.257662Z",
     "iopub.status.busy": "2023-12-08T18:55:55.257376Z",
     "iopub.status.idle": "2023-12-08T18:55:59.734575Z",
     "shell.execute_reply": "2023-12-08T18:55:59.733933Z"
    },
    "papermill": {
     "duration": 4.486733,
     "end_time": "2023-12-08T18:55:59.736575",
     "exception": false,
     "start_time": "2023-12-08T18:55:55.249842",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "train_df_c = pd.merge(train_df, cell_embs, on='cell_type', how='left')\n",
    "train_df_cm = pd.merge(train_df_c, mol_embs, on='sm_name', how='left')\n",
    "train_df_cmg = pd.merge(train_df_cm, gene_embs, on='gene', how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "5831f0a4",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:55:59.751491Z",
     "iopub.status.busy": "2023-12-08T18:55:59.751073Z",
     "iopub.status.idle": "2023-12-08T18:56:01.581739Z",
     "shell.execute_reply": "2023-12-08T18:56:01.580822Z"
    },
    "papermill": {
     "duration": 1.84025,
     "end_time": "2023-12-08T18:56:01.583788",
     "exception": false,
     "start_time": "2023-12-08T18:55:59.743538",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "test_df_c = pd.merge(test_df, cell_embs, on='cell_type', how='left')\n",
    "test_df_cm = pd.merge(test_df_c, mol_embs, on='sm_name', how='left')\n",
    "test_df_cmg = pd.merge(test_df_cm, gene_embs, on='gene', how='left')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6ad163d4",
   "metadata": {
    "papermill": {
     "duration": 0.006394,
     "end_time": "2023-12-08T18:56:01.597203",
     "exception": false,
     "start_time": "2023-12-08T18:56:01.590809",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Delete Unused Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "23c7768a",
   "metadata": {
    "papermill": {
     "duration": 0.00659,
     "end_time": "2023-12-08T18:56:01.651433",
     "exception": false,
     "start_time": "2023-12-08T18:56:01.644843",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "And collect garbage to save some memory."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "98febbec",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:56:01.666735Z",
     "iopub.status.busy": "2023-12-08T18:56:01.666027Z",
     "iopub.status.idle": "2023-12-08T18:56:02.090748Z",
     "shell.execute_reply": "2023-12-08T18:56:02.089924Z"
    },
    "papermill": {
     "duration": 0.434314,
     "end_time": "2023-12-08T18:56:02.092597",
     "exception": false,
     "start_time": "2023-12-08T18:56:01.658283",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import gc\n",
    "del train_df_c\n",
    "del train_df_cm\n",
    "del test_df_c\n",
    "del test_df_cm\n",
    "del cell_embs\n",
    "del mol_embs\n",
    "del gene_embs\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8b2b25b0",
   "metadata": {
    "papermill": {
     "duration": 0.006616,
     "end_time": "2023-12-08T18:56:02.106510",
     "exception": false,
     "start_time": "2023-12-08T18:56:02.099894",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Feature Hierarchie"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3ab3b3f6",
   "metadata": {
    "papermill": {
     "duration": 0.006566,
     "end_time": "2023-12-08T18:56:02.119895",
     "exception": false,
     "start_time": "2023-12-08T18:56:02.113329",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Convert data and check features for redundancy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "7bb096de",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:56:02.134937Z",
     "iopub.status.busy": "2023-12-08T18:56:02.134595Z",
     "iopub.status.idle": "2023-12-08T18:56:08.848589Z",
     "shell.execute_reply": "2023-12-08T18:56:08.847973Z"
    },
    "papermill": {
     "duration": 6.723365,
     "end_time": "2023-12-08T18:56:08.850122",
     "exception": false,
     "start_time": "2023-12-08T18:56:02.126757",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.55 s, sys: 2.16 s, total: 6.71 s\n",
      "Wall time: 6.71 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "to = get_train_valid_data(train_df_cmg, test_df_cmg, valid=None, random_state=random_seed)\n",
    "xs, y = to.train.xs, to.train.y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "ecb25196",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:56:08.865049Z",
     "iopub.status.busy": "2023-12-08T18:56:08.864772Z",
     "iopub.status.idle": "2023-12-08T18:56:25.361437Z",
     "shell.execute_reply": "2023-12-08T18:56:25.360371Z"
    },
    "papermill": {
     "duration": 16.511742,
     "end_time": "2023-12-08T18:56:25.368911",
     "exception": false,
     "start_time": "2023-12-08T18:56:08.857169",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 16.2 s, sys: 449 ms, total: 16.7 s\n",
      "Wall time: 16.5 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "from scipy.cluster import hierarchy\n",
    "from scipy.spatial.distance import squareform\n",
    "correlation_matrix = xs.corr()\n",
    "distances = 1 - np.abs(correlation_matrix)\n",
    "np.fill_diagonal(distances.values, 0)\n",
    "condensed_distances = squareform(distances)\n",
    "linkage_matrix = hierarchy.linkage(condensed_distances, method='complete')\n",
    "plt.figure(figsize=(10,10))\n",
    "hierarchy.dendrogram(linkage_matrix, labels=xs.columns.tolist(), orientation='left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "41c225cf",
   "metadata": {
    "papermill": {
     "duration": 0.007932,
     "end_time": "2023-12-08T18:56:25.389016",
     "exception": false,
     "start_time": "2023-12-08T18:56:25.381084",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Random Forest All Data"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8dc8a4a1",
   "metadata": {
    "papermill": {
     "duration": 0.007515,
     "end_time": "2023-12-08T18:56:25.405157",
     "exception": false,
     "start_time": "2023-12-08T18:56:25.397642",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "Finally I train a random forest using all training data, predict test scores and submit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "99a32435",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T18:56:25.426907Z",
     "iopub.status.busy": "2023-12-08T18:56:25.426198Z",
     "iopub.status.idle": "2023-12-08T19:29:58.644170Z",
     "shell.execute_reply": "2023-12-08T19:29:58.643260Z"
    },
    "papermill": {
     "duration": 2013.238236,
     "end_time": "2023-12-08T19:29:58.651664",
     "exception": false,
     "start_time": "2023-12-08T18:56:25.413428",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1h 50min 38s, sys: 42.3 s, total: 1h 51min 21s\n",
      "Wall time: 33min 33s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.864132, 1.05716, 0.8008796122295169)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "m = rf(xs, y, random_state=random_seed)\n",
    "m_rmse(m, xs, y), rmse(m.oob_prediction_, y), m.oob_score_"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c4c6e8ed",
   "metadata": {
    "papermill": {
     "duration": 0.007601,
     "end_time": "2023-12-08T19:29:58.667271",
     "exception": false,
     "start_time": "2023-12-08T19:29:58.659670",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "## Inference and Submission"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "9699f7ec",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T19:29:58.684281Z",
     "iopub.status.busy": "2023-12-08T19:29:58.683968Z",
     "iopub.status.idle": "2023-12-08T19:30:00.492608Z",
     "shell.execute_reply": "2023-12-08T19:30:00.492000Z"
    },
    "papermill": {
     "duration": 1.819191,
     "end_time": "2023-12-08T19:30:00.494276",
     "exception": false,
     "start_time": "2023-12-08T19:29:58.675085",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>cell_type</th>\n",
       "      <th>sm_name</th>\n",
       "      <th>gene</th>\n",
       "      <th>cemb0</th>\n",
       "      <th>cemb1</th>\n",
       "      <th>cemb2</th>\n",
       "      <th>cemb3</th>\n",
       "      <th>cemb4</th>\n",
       "      <th>membpca10_0</th>\n",
       "      <th>membpca10_1</th>\n",
       "      <th>...</th>\n",
       "      <th>gembpca10_0</th>\n",
       "      <th>gembpca10_1</th>\n",
       "      <th>gembpca10_2</th>\n",
       "      <th>gembpca10_3</th>\n",
       "      <th>gembpca10_4</th>\n",
       "      <th>gembpca10_5</th>\n",
       "      <th>gembpca10_6</th>\n",
       "      <th>gembpca10_7</th>\n",
       "      <th>gembpca10_8</th>\n",
       "      <th>gembpca10_9</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0.231047</td>\n",
       "      <td>0.020016</td>\n",
       "      <td>-0.090010</td>\n",
       "      <td>0.082302</td>\n",
       "      <td>-0.019138</td>\n",
       "      <td>2.524892</td>\n",
       "      <td>-2.715401</td>\n",
       "      <td>...</td>\n",
       "      <td>-5.783795</td>\n",
       "      <td>-1.162065</td>\n",
       "      <td>-5.487894</td>\n",
       "      <td>0.597049</td>\n",
       "      <td>-2.414790</td>\n",
       "      <td>1.982280</td>\n",
       "      <td>-1.08666</td>\n",
       "      <td>-1.206362</td>\n",
       "      <td>-3.901042</td>\n",
       "      <td>1.390772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0.231047</td>\n",
       "      <td>0.020016</td>\n",
       "      <td>-0.090010</td>\n",
       "      <td>0.082302</td>\n",
       "      <td>-0.019138</td>\n",
       "      <td>-1.219866</td>\n",
       "      <td>-0.877527</td>\n",
       "      <td>...</td>\n",
       "      <td>-5.783795</td>\n",
       "      <td>-1.162065</td>\n",
       "      <td>-5.487894</td>\n",
       "      <td>0.597049</td>\n",
       "      <td>-2.414790</td>\n",
       "      <td>1.982280</td>\n",
       "      <td>-1.08666</td>\n",
       "      <td>-1.206362</td>\n",
       "      <td>-3.901042</td>\n",
       "      <td>1.390772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>0.231047</td>\n",
       "      <td>0.020016</td>\n",
       "      <td>-0.090010</td>\n",
       "      <td>0.082302</td>\n",
       "      <td>-0.019138</td>\n",
       "      <td>0.025023</td>\n",
       "      <td>0.075566</td>\n",
       "      <td>...</td>\n",
       "      <td>-5.783795</td>\n",
       "      <td>-1.162065</td>\n",
       "      <td>-5.487894</td>\n",
       "      <td>0.597049</td>\n",
       "      <td>-2.414790</td>\n",
       "      <td>1.982280</td>\n",
       "      <td>-1.08666</td>\n",
       "      <td>-1.206362</td>\n",
       "      <td>-3.901042</td>\n",
       "      <td>1.390772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>0.231047</td>\n",
       "      <td>0.020016</td>\n",
       "      <td>-0.090010</td>\n",
       "      <td>0.082302</td>\n",
       "      <td>-0.019138</td>\n",
       "      <td>-1.469350</td>\n",
       "      <td>-1.851055</td>\n",
       "      <td>...</td>\n",
       "      <td>-5.783795</td>\n",
       "      <td>-1.162065</td>\n",
       "      <td>-5.487894</td>\n",
       "      <td>0.597049</td>\n",
       "      <td>-2.414790</td>\n",
       "      <td>1.982280</td>\n",
       "      <td>-1.08666</td>\n",
       "      <td>-1.206362</td>\n",
       "      <td>-3.901042</td>\n",
       "      <td>1.390772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.231047</td>\n",
       "      <td>0.020016</td>\n",
       "      <td>-0.090010</td>\n",
       "      <td>0.082302</td>\n",
       "      <td>-0.019138</td>\n",
       "      <td>0.551412</td>\n",
       "      <td>0.157118</td>\n",
       "      <td>...</td>\n",
       "      <td>-5.783795</td>\n",
       "      <td>-1.162065</td>\n",
       "      <td>-5.487894</td>\n",
       "      <td>0.597049</td>\n",
       "      <td>-2.414790</td>\n",
       "      <td>1.982280</td>\n",
       "      <td>-1.08666</td>\n",
       "      <td>-1.206362</td>\n",
       "      <td>-3.901042</td>\n",
       "      <td>1.390772</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4643800</th>\n",
       "      <td>2</td>\n",
       "      <td>142</td>\n",
       "      <td>18211</td>\n",
       "      <td>-0.130559</td>\n",
       "      <td>0.138034</td>\n",
       "      <td>0.096609</td>\n",
       "      <td>0.036904</td>\n",
       "      <td>0.094145</td>\n",
       "      <td>-1.627801</td>\n",
       "      <td>2.048937</td>\n",
       "      <td>...</td>\n",
       "      <td>-7.473646</td>\n",
       "      <td>13.909621</td>\n",
       "      <td>3.644948</td>\n",
       "      <td>-3.208391</td>\n",
       "      <td>0.624002</td>\n",
       "      <td>1.707114</td>\n",
       "      <td>-1.91051</td>\n",
       "      <td>-1.623606</td>\n",
       "      <td>-2.552454</td>\n",
       "      <td>-1.376674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4643801</th>\n",
       "      <td>2</td>\n",
       "      <td>143</td>\n",
       "      <td>18211</td>\n",
       "      <td>-0.130559</td>\n",
       "      <td>0.138034</td>\n",
       "      <td>0.096609</td>\n",
       "      <td>0.036904</td>\n",
       "      <td>0.094145</td>\n",
       "      <td>-0.976956</td>\n",
       "      <td>-0.495422</td>\n",
       "      <td>...</td>\n",
       "      <td>-7.473646</td>\n",
       "      <td>13.909621</td>\n",
       "      <td>3.644948</td>\n",
       "      <td>-3.208391</td>\n",
       "      <td>0.624002</td>\n",
       "      <td>1.707114</td>\n",
       "      <td>-1.91051</td>\n",
       "      <td>-1.623606</td>\n",
       "      <td>-2.552454</td>\n",
       "      <td>-1.376674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4643802</th>\n",
       "      <td>2</td>\n",
       "      <td>144</td>\n",
       "      <td>18211</td>\n",
       "      <td>-0.130559</td>\n",
       "      <td>0.138034</td>\n",
       "      <td>0.096609</td>\n",
       "      <td>0.036904</td>\n",
       "      <td>0.094145</td>\n",
       "      <td>-1.961592</td>\n",
       "      <td>-0.395690</td>\n",
       "      <td>...</td>\n",
       "      <td>-7.473646</td>\n",
       "      <td>13.909621</td>\n",
       "      <td>3.644948</td>\n",
       "      <td>-3.208391</td>\n",
       "      <td>0.624002</td>\n",
       "      <td>1.707114</td>\n",
       "      <td>-1.91051</td>\n",
       "      <td>-1.623606</td>\n",
       "      <td>-2.552454</td>\n",
       "      <td>-1.376674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4643803</th>\n",
       "      <td>2</td>\n",
       "      <td>145</td>\n",
       "      <td>18211</td>\n",
       "      <td>-0.130559</td>\n",
       "      <td>0.138034</td>\n",
       "      <td>0.096609</td>\n",
       "      <td>0.036904</td>\n",
       "      <td>0.094145</td>\n",
       "      <td>0.528172</td>\n",
       "      <td>2.419617</td>\n",
       "      <td>...</td>\n",
       "      <td>-7.473646</td>\n",
       "      <td>13.909621</td>\n",
       "      <td>3.644948</td>\n",
       "      <td>-3.208391</td>\n",
       "      <td>0.624002</td>\n",
       "      <td>1.707114</td>\n",
       "      <td>-1.91051</td>\n",
       "      <td>-1.623606</td>\n",
       "      <td>-2.552454</td>\n",
       "      <td>-1.376674</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4643804</th>\n",
       "      <td>2</td>\n",
       "      <td>146</td>\n",
       "      <td>18211</td>\n",
       "      <td>-0.130559</td>\n",
       "      <td>0.138034</td>\n",
       "      <td>0.096609</td>\n",
       "      <td>0.036904</td>\n",
       "      <td>0.094145</td>\n",
       "      <td>-1.346407</td>\n",
       "      <td>0.516794</td>\n",
       "      <td>...</td>\n",
       "      <td>-7.473646</td>\n",
       "      <td>13.909621</td>\n",
       "      <td>3.644948</td>\n",
       "      <td>-3.208391</td>\n",
       "      <td>0.624002</td>\n",
       "      <td>1.707114</td>\n",
       "      <td>-1.91051</td>\n",
       "      <td>-1.623606</td>\n",
       "      <td>-2.552454</td>\n",
       "      <td>-1.376674</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4643805 rows × 28 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         cell_type  sm_name   gene     cemb0     cemb1     cemb2     cemb3  \\\n",
       "0                1        1      1  0.231047  0.020016 -0.090010  0.082302   \n",
       "1                1        2      1  0.231047  0.020016 -0.090010  0.082302   \n",
       "2                1        3      1  0.231047  0.020016 -0.090010  0.082302   \n",
       "3                1        4      1  0.231047  0.020016 -0.090010  0.082302   \n",
       "4                1        5      1  0.231047  0.020016 -0.090010  0.082302   \n",
       "...            ...      ...    ...       ...       ...       ...       ...   \n",
       "4643800          2      142  18211 -0.130559  0.138034  0.096609  0.036904   \n",
       "4643801          2      143  18211 -0.130559  0.138034  0.096609  0.036904   \n",
       "4643802          2      144  18211 -0.130559  0.138034  0.096609  0.036904   \n",
       "4643803          2      145  18211 -0.130559  0.138034  0.096609  0.036904   \n",
       "4643804          2      146  18211 -0.130559  0.138034  0.096609  0.036904   \n",
       "\n",
       "            cemb4  membpca10_0  membpca10_1  ...  gembpca10_0  gembpca10_1  \\\n",
       "0       -0.019138     2.524892    -2.715401  ...    -5.783795    -1.162065   \n",
       "1       -0.019138    -1.219866    -0.877527  ...    -5.783795    -1.162065   \n",
       "2       -0.019138     0.025023     0.075566  ...    -5.783795    -1.162065   \n",
       "3       -0.019138    -1.469350    -1.851055  ...    -5.783795    -1.162065   \n",
       "4       -0.019138     0.551412     0.157118  ...    -5.783795    -1.162065   \n",
       "...           ...          ...          ...  ...          ...          ...   \n",
       "4643800  0.094145    -1.627801     2.048937  ...    -7.473646    13.909621   \n",
       "4643801  0.094145    -0.976956    -0.495422  ...    -7.473646    13.909621   \n",
       "4643802  0.094145    -1.961592    -0.395690  ...    -7.473646    13.909621   \n",
       "4643803  0.094145     0.528172     2.419617  ...    -7.473646    13.909621   \n",
       "4643804  0.094145    -1.346407     0.516794  ...    -7.473646    13.909621   \n",
       "\n",
       "         gembpca10_2  gembpca10_3  gembpca10_4  gembpca10_5  gembpca10_6  \\\n",
       "0          -5.487894     0.597049    -2.414790     1.982280     -1.08666   \n",
       "1          -5.487894     0.597049    -2.414790     1.982280     -1.08666   \n",
       "2          -5.487894     0.597049    -2.414790     1.982280     -1.08666   \n",
       "3          -5.487894     0.597049    -2.414790     1.982280     -1.08666   \n",
       "4          -5.487894     0.597049    -2.414790     1.982280     -1.08666   \n",
       "...              ...          ...          ...          ...          ...   \n",
       "4643800     3.644948    -3.208391     0.624002     1.707114     -1.91051   \n",
       "4643801     3.644948    -3.208391     0.624002     1.707114     -1.91051   \n",
       "4643802     3.644948    -3.208391     0.624002     1.707114     -1.91051   \n",
       "4643803     3.644948    -3.208391     0.624002     1.707114     -1.91051   \n",
       "4643804     3.644948    -3.208391     0.624002     1.707114     -1.91051   \n",
       "\n",
       "         gembpca10_7  gembpca10_8  gembpca10_9  \n",
       "0          -1.206362    -3.901042     1.390772  \n",
       "1          -1.206362    -3.901042     1.390772  \n",
       "2          -1.206362    -3.901042     1.390772  \n",
       "3          -1.206362    -3.901042     1.390772  \n",
       "4          -1.206362    -3.901042     1.390772  \n",
       "...              ...          ...          ...  \n",
       "4643800    -1.623606    -2.552454    -1.376674  \n",
       "4643801    -1.623606    -2.552454    -1.376674  \n",
       "4643802    -1.623606    -2.552454    -1.376674  \n",
       "4643803    -1.623606    -2.552454    -1.376674  \n",
       "4643804    -1.623606    -2.552454    -1.376674  \n",
       "\n",
       "[4643805 rows x 28 columns]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_to = to.new(test_df_cmg)\n",
    "test_to.process()\n",
    "test_xs = test_to.xs\n",
    "test_xs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "ccb5c25d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T19:30:00.513010Z",
     "iopub.status.busy": "2023-12-08T19:30:00.512524Z",
     "iopub.status.idle": "2023-12-08T19:30:55.244338Z",
     "shell.execute_reply": "2023-12-08T19:30:55.243426Z"
    },
    "papermill": {
     "duration": 54.751199,
     "end_time": "2023-12-08T19:30:55.254240",
     "exception": false,
     "start_time": "2023-12-08T19:30:00.503041",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([ 0.2320,  0.1719,  0.4196,  ..., -0.0962, -0.0292, -0.0715])"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds = tensor(m.predict(test_xs))\n",
    "preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "be99d2ea",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T19:30:55.271939Z",
     "iopub.status.busy": "2023-12-08T19:30:55.271478Z",
     "iopub.status.idle": "2023-12-08T19:30:55.278639Z",
     "shell.execute_reply": "2023-12-08T19:30:55.277927Z"
    },
    "papermill": {
     "duration": 0.017753,
     "end_time": "2023-12-08T19:30:55.280118",
     "exception": false,
     "start_time": "2023-12-08T19:30:55.262365",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(tensor(-18.3912), tensor(29.9948))"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "preds.min(), preds.max()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "657c06d7",
   "metadata": {
    "papermill": {
     "duration": 0.009175,
     "end_time": "2023-12-08T19:30:55.297720",
     "exception": false,
     "start_time": "2023-12-08T19:30:55.288545",
     "status": "completed"
    },
    "tags": []
   },
   "source": [
    "In the following step I reshape the predictions back into a 255 x 18211 tensor for submission:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "d921d69a",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T19:30:55.315923Z",
     "iopub.status.busy": "2023-12-08T19:30:55.315365Z",
     "iopub.status.idle": "2023-12-08T19:30:55.379030Z",
     "shell.execute_reply": "2023-12-08T19:30:55.378149Z"
    },
    "papermill": {
     "duration": 0.075472,
     "end_time": "2023-12-08T19:30:55.381427",
     "exception": false,
     "start_time": "2023-12-08T19:30:55.305955",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>A1BG</th>\n",
       "      <th>A1BG-AS1</th>\n",
       "      <th>A2M</th>\n",
       "      <th>A2M-AS1</th>\n",
       "      <th>A2MP1</th>\n",
       "      <th>A4GALT</th>\n",
       "      <th>AAAS</th>\n",
       "      <th>AACS</th>\n",
       "      <th>AAGAB</th>\n",
       "      <th>AAK1</th>\n",
       "      <th>...</th>\n",
       "      <th>ZUP1</th>\n",
       "      <th>ZW10</th>\n",
       "      <th>ZWILCH</th>\n",
       "      <th>ZWINT</th>\n",
       "      <th>ZXDA</th>\n",
       "      <th>ZXDB</th>\n",
       "      <th>ZXDC</th>\n",
       "      <th>ZYG11B</th>\n",
       "      <th>ZYX</th>\n",
       "      <th>ZZEF1</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.232016</td>\n",
       "      <td>0.110668</td>\n",
       "      <td>0.440268</td>\n",
       "      <td>0.609447</td>\n",
       "      <td>1.480283</td>\n",
       "      <td>1.181870</td>\n",
       "      <td>0.125640</td>\n",
       "      <td>0.108322</td>\n",
       "      <td>-0.123958</td>\n",
       "      <td>0.358815</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.321615</td>\n",
       "      <td>-0.068769</td>\n",
       "      <td>0.195414</td>\n",
       "      <td>0.255237</td>\n",
       "      <td>0.236392</td>\n",
       "      <td>0.174145</td>\n",
       "      <td>0.248805</td>\n",
       "      <td>0.415817</td>\n",
       "      <td>0.088075</td>\n",
       "      <td>0.071194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.171852</td>\n",
       "      <td>0.075148</td>\n",
       "      <td>0.149117</td>\n",
       "      <td>0.214129</td>\n",
       "      <td>1.191716</td>\n",
       "      <td>1.263651</td>\n",
       "      <td>-0.004535</td>\n",
       "      <td>0.158479</td>\n",
       "      <td>0.107841</td>\n",
       "      <td>-0.016537</td>\n",
       "      <td>...</td>\n",
       "      <td>0.113699</td>\n",
       "      <td>0.100706</td>\n",
       "      <td>-0.021103</td>\n",
       "      <td>0.153068</td>\n",
       "      <td>0.192868</td>\n",
       "      <td>0.048186</td>\n",
       "      <td>0.142199</td>\n",
       "      <td>0.146898</td>\n",
       "      <td>0.150609</td>\n",
       "      <td>-0.053352</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.419615</td>\n",
       "      <td>0.099909</td>\n",
       "      <td>0.041278</td>\n",
       "      <td>0.107889</td>\n",
       "      <td>1.568029</td>\n",
       "      <td>2.845965</td>\n",
       "      <td>0.050505</td>\n",
       "      <td>0.323187</td>\n",
       "      <td>0.143395</td>\n",
       "      <td>0.005585</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.080905</td>\n",
       "      <td>0.002012</td>\n",
       "      <td>0.061581</td>\n",
       "      <td>0.378854</td>\n",
       "      <td>0.183387</td>\n",
       "      <td>0.273359</td>\n",
       "      <td>0.246448</td>\n",
       "      <td>0.117373</td>\n",
       "      <td>-0.003684</td>\n",
       "      <td>-0.096323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.039393</td>\n",
       "      <td>0.221065</td>\n",
       "      <td>0.108384</td>\n",
       "      <td>0.289663</td>\n",
       "      <td>0.887017</td>\n",
       "      <td>0.646559</td>\n",
       "      <td>-0.087853</td>\n",
       "      <td>0.153870</td>\n",
       "      <td>0.101997</td>\n",
       "      <td>0.334735</td>\n",
       "      <td>...</td>\n",
       "      <td>0.036526</td>\n",
       "      <td>0.116040</td>\n",
       "      <td>0.047634</td>\n",
       "      <td>0.152395</td>\n",
       "      <td>-0.016641</td>\n",
       "      <td>-0.012576</td>\n",
       "      <td>0.086604</td>\n",
       "      <td>0.343730</td>\n",
       "      <td>0.100656</td>\n",
       "      <td>0.057325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.326365</td>\n",
       "      <td>0.070912</td>\n",
       "      <td>0.262907</td>\n",
       "      <td>0.225060</td>\n",
       "      <td>1.144794</td>\n",
       "      <td>1.209496</td>\n",
       "      <td>0.082896</td>\n",
       "      <td>0.187761</td>\n",
       "      <td>0.091944</td>\n",
       "      <td>0.366254</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.190770</td>\n",
       "      <td>0.121339</td>\n",
       "      <td>0.090370</td>\n",
       "      <td>0.181342</td>\n",
       "      <td>0.104310</td>\n",
       "      <td>0.143238</td>\n",
       "      <td>0.246384</td>\n",
       "      <td>0.407660</td>\n",
       "      <td>0.146642</td>\n",
       "      <td>-0.037769</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>250</th>\n",
       "      <td>0.061888</td>\n",
       "      <td>-0.115301</td>\n",
       "      <td>-0.171353</td>\n",
       "      <td>-0.160975</td>\n",
       "      <td>0.083676</td>\n",
       "      <td>-0.050931</td>\n",
       "      <td>0.016860</td>\n",
       "      <td>-0.073320</td>\n",
       "      <td>-0.043505</td>\n",
       "      <td>0.237127</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.081356</td>\n",
       "      <td>-0.184846</td>\n",
       "      <td>-0.305815</td>\n",
       "      <td>-0.091784</td>\n",
       "      <td>-0.009122</td>\n",
       "      <td>-0.069461</td>\n",
       "      <td>-0.117386</td>\n",
       "      <td>0.022502</td>\n",
       "      <td>-0.068701</td>\n",
       "      <td>0.009319</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>251</th>\n",
       "      <td>0.089803</td>\n",
       "      <td>-0.078957</td>\n",
       "      <td>0.097619</td>\n",
       "      <td>0.335744</td>\n",
       "      <td>0.645069</td>\n",
       "      <td>0.471046</td>\n",
       "      <td>0.003653</td>\n",
       "      <td>0.122890</td>\n",
       "      <td>0.184567</td>\n",
       "      <td>0.221382</td>\n",
       "      <td>...</td>\n",
       "      <td>0.101355</td>\n",
       "      <td>0.039257</td>\n",
       "      <td>-0.402044</td>\n",
       "      <td>-0.019977</td>\n",
       "      <td>0.243823</td>\n",
       "      <td>0.035454</td>\n",
       "      <td>-0.021406</td>\n",
       "      <td>0.079556</td>\n",
       "      <td>-0.206459</td>\n",
       "      <td>-0.084462</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>252</th>\n",
       "      <td>-0.075844</td>\n",
       "      <td>-0.032259</td>\n",
       "      <td>-0.096995</td>\n",
       "      <td>-0.140546</td>\n",
       "      <td>0.007538</td>\n",
       "      <td>-0.099034</td>\n",
       "      <td>-0.043268</td>\n",
       "      <td>-0.063726</td>\n",
       "      <td>0.072114</td>\n",
       "      <td>0.083702</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.100846</td>\n",
       "      <td>-0.152247</td>\n",
       "      <td>-0.136672</td>\n",
       "      <td>0.003205</td>\n",
       "      <td>-0.073530</td>\n",
       "      <td>-0.033837</td>\n",
       "      <td>0.000309</td>\n",
       "      <td>0.069686</td>\n",
       "      <td>0.030961</td>\n",
       "      <td>-0.096164</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>253</th>\n",
       "      <td>1.081692</td>\n",
       "      <td>0.673840</td>\n",
       "      <td>0.639014</td>\n",
       "      <td>0.589074</td>\n",
       "      <td>2.015911</td>\n",
       "      <td>2.339699</td>\n",
       "      <td>0.511117</td>\n",
       "      <td>0.932566</td>\n",
       "      <td>0.579995</td>\n",
       "      <td>0.595701</td>\n",
       "      <td>...</td>\n",
       "      <td>0.115555</td>\n",
       "      <td>0.656881</td>\n",
       "      <td>-0.010994</td>\n",
       "      <td>0.752008</td>\n",
       "      <td>0.695284</td>\n",
       "      <td>0.754038</td>\n",
       "      <td>0.251581</td>\n",
       "      <td>0.378218</td>\n",
       "      <td>0.128816</td>\n",
       "      <td>-0.029185</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>254</th>\n",
       "      <td>0.045273</td>\n",
       "      <td>0.111711</td>\n",
       "      <td>-0.056743</td>\n",
       "      <td>-0.047229</td>\n",
       "      <td>0.258056</td>\n",
       "      <td>0.147763</td>\n",
       "      <td>-0.014319</td>\n",
       "      <td>0.221553</td>\n",
       "      <td>0.102350</td>\n",
       "      <td>0.097696</td>\n",
       "      <td>...</td>\n",
       "      <td>0.161723</td>\n",
       "      <td>0.101919</td>\n",
       "      <td>-0.351171</td>\n",
       "      <td>0.011027</td>\n",
       "      <td>0.126013</td>\n",
       "      <td>0.072658</td>\n",
       "      <td>-0.031409</td>\n",
       "      <td>-0.072564</td>\n",
       "      <td>-0.133203</td>\n",
       "      <td>-0.071533</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>255 rows × 18211 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         A1BG  A1BG-AS1       A2M   A2M-AS1     A2MP1    A4GALT      AAAS  \\\n",
       "id                                                                          \n",
       "0    0.232016  0.110668  0.440268  0.609447  1.480283  1.181870  0.125640   \n",
       "1    0.171852  0.075148  0.149117  0.214129  1.191716  1.263651 -0.004535   \n",
       "2    0.419615  0.099909  0.041278  0.107889  1.568029  2.845965  0.050505   \n",
       "3    0.039393  0.221065  0.108384  0.289663  0.887017  0.646559 -0.087853   \n",
       "4    0.326365  0.070912  0.262907  0.225060  1.144794  1.209496  0.082896   \n",
       "..        ...       ...       ...       ...       ...       ...       ...   \n",
       "250  0.061888 -0.115301 -0.171353 -0.160975  0.083676 -0.050931  0.016860   \n",
       "251  0.089803 -0.078957  0.097619  0.335744  0.645069  0.471046  0.003653   \n",
       "252 -0.075844 -0.032259 -0.096995 -0.140546  0.007538 -0.099034 -0.043268   \n",
       "253  1.081692  0.673840  0.639014  0.589074  2.015911  2.339699  0.511117   \n",
       "254  0.045273  0.111711 -0.056743 -0.047229  0.258056  0.147763 -0.014319   \n",
       "\n",
       "         AACS     AAGAB      AAK1  ...      ZUP1      ZW10    ZWILCH  \\\n",
       "id                                 ...                                 \n",
       "0    0.108322 -0.123958  0.358815  ... -0.321615 -0.068769  0.195414   \n",
       "1    0.158479  0.107841 -0.016537  ...  0.113699  0.100706 -0.021103   \n",
       "2    0.323187  0.143395  0.005585  ... -0.080905  0.002012  0.061581   \n",
       "3    0.153870  0.101997  0.334735  ...  0.036526  0.116040  0.047634   \n",
       "4    0.187761  0.091944  0.366254  ... -0.190770  0.121339  0.090370   \n",
       "..        ...       ...       ...  ...       ...       ...       ...   \n",
       "250 -0.073320 -0.043505  0.237127  ... -0.081356 -0.184846 -0.305815   \n",
       "251  0.122890  0.184567  0.221382  ...  0.101355  0.039257 -0.402044   \n",
       "252 -0.063726  0.072114  0.083702  ... -0.100846 -0.152247 -0.136672   \n",
       "253  0.932566  0.579995  0.595701  ...  0.115555  0.656881 -0.010994   \n",
       "254  0.221553  0.102350  0.097696  ...  0.161723  0.101919 -0.351171   \n",
       "\n",
       "        ZWINT      ZXDA      ZXDB      ZXDC    ZYG11B       ZYX     ZZEF1  \n",
       "id                                                                         \n",
       "0    0.255237  0.236392  0.174145  0.248805  0.415817  0.088075  0.071194  \n",
       "1    0.153068  0.192868  0.048186  0.142199  0.146898  0.150609 -0.053352  \n",
       "2    0.378854  0.183387  0.273359  0.246448  0.117373 -0.003684 -0.096323  \n",
       "3    0.152395 -0.016641 -0.012576  0.086604  0.343730  0.100656  0.057325  \n",
       "4    0.181342  0.104310  0.143238  0.246384  0.407660  0.146642 -0.037769  \n",
       "..        ...       ...       ...       ...       ...       ...       ...  \n",
       "250 -0.091784 -0.009122 -0.069461 -0.117386  0.022502 -0.068701  0.009319  \n",
       "251 -0.019977  0.243823  0.035454 -0.021406  0.079556 -0.206459 -0.084462  \n",
       "252  0.003205 -0.073530 -0.033837  0.000309  0.069686  0.030961 -0.096164  \n",
       "253  0.752008  0.695284  0.754038  0.251581  0.378218  0.128816 -0.029185  \n",
       "254  0.011027  0.126013  0.072658 -0.031409 -0.072564 -0.133203 -0.071533  \n",
       "\n",
       "[255 rows x 18211 columns]"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "to_submit = preds.view(18211, -1).t().numpy()\n",
    "submit = pd.DataFrame(to_submit, columns=df_de_train.iloc[:,5:].columns)\n",
    "submit.index.name = 'id'\n",
    "submit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "fce1186b",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-12-08T19:30:55.403382Z",
     "iopub.status.busy": "2023-12-08T19:30:55.403033Z",
     "iopub.status.idle": "2023-12-08T19:30:59.875846Z",
     "shell.execute_reply": "2023-12-08T19:30:59.875102Z"
    },
    "papermill": {
     "duration": 4.485237,
     "end_time": "2023-12-08T19:30:59.878093",
     "exception": false,
     "start_time": "2023-12-08T19:30:55.392856",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "submit.to_csv('submission.csv')"
   ]
  }
 ],
 "metadata": {
  "kaggle": {
   "accelerator": "none",
   "dataSources": [
    {
     "databundleVersionId": 7010844,
     "sourceId": 59094,
     "sourceType": "competition"
    },
    {
     "datasetId": 4109071,
     "sourceId": 7123483,
     "sourceType": "datasetVersion"
    }
   ],
   "dockerImageVersionId": 30587,
   "isGpuEnabled": false,
   "isInternetEnabled": true,
   "language": "python",
   "sourceType": "notebook"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 2125.009428,
   "end_time": "2023-12-08T19:31:02.026246",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2023-12-08T18:55:37.016818",
   "version": "2.4.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}