Analyzing intra-year daily average temperatures

We are going to have a look at the temperature variation within a year by converting dates to the corresponding day of the year in numbers. This number is between 1 and 366, where 1 corresponds to January 1st and 365 (or 366) corresponds to December 31st. Perform the following steps to analyze the intra-year daily average temperature:

1. Initialize arrays for the range 1-366 with averages initialized to zeros:

   rng = np.arange(1, 366)
   avgs = np.zeros(365)
   avgs2 = np.zeros(365)

2. Calculate averages by the day of the year before and after a cutoff point:

   for i in rng: 
      indices = np.where(days[:cutoff] == i)
      avgs[i-1] = temp[indices].mean()
      indices = np.where(days[cutoff+1:] == i)
      avgs2[i-1] = temp[indices].mean()

3. Fit the averages before the cutoff point to a quadratic polynomial (just a first-order approximation):

   poly = np.polyfit(rng, avgs, 2)
   print poly

   The following polynomial coefficients in descending power are printed:

   [ -4.91329859e-04...