df_cars['mpg'] = df_cars['mpg'].fillna((df_cars.mpg).mean())

best_r2 = 0
best_n = 0;
for n_value in range (1, 16):
    ls = np.polyfit(df_cars.mpg, df_cars.wt, n_value)
    p = np.poly1d(ls) # Allows for operations on polynomials for calculation in graph
    xp = np.linspace(df_cars.mpg.min(), df_cars.mpg.max(), 100)
    results = []
    for x_coord in df_cars.mpg:
        results.append(p(x_coord))

    rsquared = metrics.r2_score(df_cars.wt.values, results)
    if rsquared > best_r2:
        best_r2 = rsquared
        best_n = n_value

print(f'n of {best_n}, gives bests r2 =', round(rsquared, 5))