As a first step, start with who_raw and create a new data frame, who2, with the following properties: All the 'new...' columns of who_raw become values of a single variable, case_type. Store the counts associated with each case_type value as a new variable called 'count'. Remove the iso2 and iso3 columns, since they are redundant with country (which you should keep!). Keep the year column as a variable. Remove all not-a-number (NaN) counts. Hint: You can test for a NaN using Python's math. isnan(). Convert the counts to integers. (Because of the presence of NaNs, the counts will be otherwise be treated as floating-point values, which is undesirable since you do not expect to see non-integer counts.)