Sendy Logistics Challenge
$7,000 USD
Predict the estimated time of arrival (ETA) for motorbike deliveries in Nairobi
1178 data scientists enrolled, 431 on the leaderboard
23 August 2019—26 November 2019
Is this bike or Airplane
published 7 Sep 2019, 14:06

There is some rows where time_from_pickup_to_arrival doesn't make much sense compared to the distance that the rider shoud cross. For example we can find a distance of 14 Km with a corresponding time_from_pickup_to_arrival equals 1second. What is the meaning of this case (cancellation of request, a disagreement, the driver didn't find the client, the driver didn't show up .... )

I guess its email!!!......probably one of those outliers though

Lol. Very funny subject...must be an airplane. Hehe

But very valid observation Mohamed. Whilst we can treat or remove this anomalies/out;iers, the greatest concern is if the hidden evaluation file is free from such errors. It will be pointless to model for an error.

My ask for Zindi is to please review the times to ensure they are valid and free oferrors. A subtraction of delivery and pick up times will point out some of these erros inherent.

My guess is that the rider forgot to enter that they picked up the item in the app, didn't realize it until they got to the destination, then quickly hit the "Pickup Complete" and "Dropoff Complete" buttons.

In my working practice was such situations. This errors produced by drivers, who clicked in app not in right time. I suppose that this kind of errors will be in train and in test dataset (I think data is consistent).

So funny... Probably an

Hello, these are definitely bikes.

Some entries have short travel times, this could be because the rider forgot to enter the time they picked up the parcel so quickly added it before they delivered the parcel.

Is there any chance this errors are in the test set? If that is the case, then it'll be pointless modelling for it