Insights

I found peaks in the demand, at 2 am, and 9 pm, having delivery guys ready to fullfill deliveries at such hour could improve the customer experience. A minority of the orders to Restaurant’s have good rates, I should investigate the causes, and consequently act.Exploring the rates given to delivery boys with at least 7 orders delivered I found a lookalike Gaussian distribution, with a median of 3.1; besides, a percentage had a bad performance in the span covered.

Further exploration requires more data, in obsevations and variables, e.g. cliente id.

Overview

The dataset contains 10000 rows with 11 variables, after removing some rows with dates in the amount variable we have 9981 records. The deliveries reported started the 05-15-2017 and finished the 07-17-2017.

Deliveries per day

The daily deliveries per type are shown in the next plot; the types Express and Restaurant had similar deliveries, with Super deliveries having less than half of such activity.

Deliveries per hour

The following plot displays the total deliveries per hour of the day. There are two peaks of activity, at 2 am, and at 9 pm; having delivery guys ready to fullfill deliveries at those hours could improve the customer experience.

Rate of orders

The next plot shows the user’s rate to the order. Express have greater proportion of good rates, in the middle he encounter the Super type with mixed rates, and finally we have the Restaurant types reporting more bad rates than good ones.

I would suggest to further investigate the reasons of the bad rates for the Restaurant orders.

Performance of delivery boys

In the sample shared, the delivery boys have the next distribution of orders delivered; it roughly looks like a Gaussian distribution.

The orders delivered goes from 0 to 21, I will obtain the average rate of those with at least 7 orders.

The mean rate is 3.1, with a maximum mean rate of 4.4; however, there are also bad performers, I would suggest investigate the delivery boys with a low average rate.