iX Mobile Banking Prediction Challenge
Helping Africa
$3 495 USD
Challenge completed over 4 years ago
Prediction
340 joined
103 active
Start
May 17, 21
Close
Jun 13, 21
Reveal
Jun 13, 21
Unofficial Winning Solution [ 0.5072 on Private LB ]
Notebooks · 14 Jun 2021, 05:54 · edited 1 day later · 3
Hello Zindians ,
Here is a repository that contains Solution Code on how to easily reach 1st Place in this Challenge.
Repository Link : https://github.com/ASSAZZIN-01/iX-Mobile-Banking-Prediction-Challenge .
--------------------------------------------------------------------------
Solution Code Pipeline
1. Data Processing :
1.1 Remove Columns with more than 80% Nans
1.2 Fillna FQ Features with -1
1.3 Fillna Age With Mean Age
2. Modeling : I trained an LGBM Model on each region separately so this will help our model to generalize well
--------------------------------------------------------------------------
Don't Forget to star the Repository, So I'll keep sharing my Solution Code
Thank you @ASSAZZIN for sharing your solution. I like how you approached this competition.
How did you get LGBM parameters? Did you use gridsearch, randomsearch or optuna to tune the LGBM or you already knew those are the parameters would work in this competition.
Usually, I tune Manually the HyperParameters.
For more details You can refer to this documentation : LGBM Parameters Tuning
Thank you @ASSAZZIN