Data-Days-2020

February and March of 2020. Sharif University of Technology, Tehran, Iran.

Ranked the 1st among 121 teams.

1153 Contestants| 121 Teams
86 Universities

Introduction

The 2020 Data Days Competition held at Sharif university of technology. The website link The link to the documentation The link to the dataset You are given a dataset of 9 csv files and you are expected to sort adIds that are most likely to be clicked on by a certain user. The problem and the datasets are similar to this kaggle competiton.

You can chceck the jupyter notebook and see the algorithms and preprocessing we had done.

Our team (March 2020):

Team Pandas
Milad Soltany --- Linkedin, Github , Email
Abbas Omidi --- Linkedin, Github, Email
Amirhossein Heydarian --- Linkedin, Github, Email

feel free to reach out to us.

What we did.

We tried out many different models and algorithms.
What worked the best was random forest with 400 estimators and max_depth of 15, you can check the preproccesings and features we extracted.

What you find in this repo

The final notebook which consists of all the codes.

Photos of work results

An Example of Statistical Analysis of the Number of Clicks Based on Various Parameters :

View of One of the Datasets Used to Teach the Model:

A Picture of the Importance of Our Own Features Compared to Pre-Existing Features in the Dataset:

Best.
Team Pandas.