Conclusions
Consulting a t-test with a t-stat of 5.11 and an significance level of 0.05, you would fail to reject the null if you had only 1 degree of freedom. However, our degrees of freedom is found using the Welch-Satterthwaite Equation and is extremely high (~625,000). Intuitively, you can reject the null at decreasing t stat levels as the degrees of freedom increases. In addition, by running the independent t-test through scipy.stats, we get a p value of 0.00000. Our two groups are therefore significantly different
However, our null hypothesis states that young people have smaller than or equal trip distances to old people. By just looking at the means of each trip distance, we can see that old people in fact travel longer distances. We therefore fail to reject the null.
This analysis can be improved by testing over a combination of months, in order to get rid of the effect of seasonality. We can also remove outliers in case those had any impact on the mean. Finally, we can also use the Google API to get a more accurate distance traveled by bike, instead of using the simple Haversine formula described above.