fixes to 3.*

notebooks/1.4_solution.md

@@ -0,0 +1,44 @@
+This is one possible solution to the exercise contained at the end of notebook "1.4 Skills Pandas":
+
+```python
+​
+import pandas as pd
+​
+# Read the data
+csv_path = "https://dataspace.princeton.edu/bitstream/88435/dsp01jm214s28p/2/SCoData_books_v1.2_2022-01.csv"
+# yes, in pandas you can read a csv file directly from a URL without having to
+# download it manually first. 
+scodata = pd.read_csv(csv_path)
+​
+# How many records?
+print(scodata.shape[0])
+​
+# Select columns
+scodata_copy = scodata[['uri','format','borrow_count']]
+​
+# Drop NaN values in `format` column
+scodata_copy = scodata_copy.dropna(subset=['format'])
+print(scodata_copy.shape[0])
+​
+# Most borrowed
+# alternative method is to use `scodata.borrow_count.max()`
+times_most_borrowed = sorted(
+    scodata.borrow_count.value_counts().keys().to_list()
+)[-1]
+​
+# Format(s) of the most borrowed documents
+scodata_copy[
+    scodata_copy.borrow_count==times_most_borrowed
+].format.to_list()
+​
+# Least borrowed
+# alternative method is to use `scodata.borrow_count.min()`
+times_least_borrowed = sorted(
+    scodata.borrow_count.value_counts().keys().to_list()
+)[0]
+​
+# Format(s) of the least borrowed documents
+scodata_copy[
+    scodata_copy.borrow_count==times_least_borrowed
+].format.unique().tolist()
+```