# My approach (4.22 CV / 3.704 public LB)
**URL:** https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351
**Category:** ADIA Lab
**Created:** [August 18, 2023, 9:28am UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351 "2023-08-18T09:28:28Z")
**Posts on this page:** 7
**Page:** 1
### Author:  [@salty-francisco](https://forum.crunchdao.com/u/salty-francisco)
#### Post date: [August 18, 2023, 9:28am UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/1 "2023-08-18T09:28:28Z")
**Approach**
I’d like to detail my approach and share the validation results.
All validations results are calculated as an average from moon 50 to moon 268.
I used four types of models:
- LightGBM (CV 3.74) (100 features) (loss rmse)
- ExtraTreesRegressor (CV 3.65) (100 features) (loss squared\_error)
- Catboost (CV 3.68) (89 features) (loss rmse)
- NN (CV 3.56) (89 features) (loss mse)
I selected the features with forward feature selection.
**Validation and Results:**
Through stacking these models I got CV 4.22 (LB 3.704).
**Target transformation**
np.log1p(target)
**Extra features**
Number of IDs in the previous moons
**CV-LB consistency**
I found that big improvements were reflected in LB but sometimes small improvements were not reflected due to certain degree of randomness that exists
**Historical data and retraining**
I found that just using last 10 moons for training gives the best results in my case, so I put 1 retrain every moon and the training data is the last 10 moons.
**Observation on Randomness:**
During the course of the competition, my best public submission (4.15 LB) was obtained inadvertently due to a bug in the code. This discovery sheds light on the presence of a certain degree of randomness inherent in the competition.
---
### Author:  [@subtle-rajat](https://forum.crunchdao.com/u/subtle-rajat)
#### Post date: [August 18, 2023, 4:46pm UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/2 "2023-08-18T16:46:33Z")
What was the train frequency set?
---
### Author:  [@salty-francisco](https://forum.crunchdao.com/u/salty-francisco)
#### Post date: [August 18, 2023, 5:18pm UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/3 "2023-08-18T17:18:31Z")
I train it every moon because my training time is short.
---
### Author:  [@narrow-oskar](https://forum.crunchdao.com/u/narrow-oskar)
#### Post date: [August 21, 2023, 12:31pm UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/4 "2023-08-21T12:31:22Z")
The CV scores you presented were calculated using a rolling-origin CV (retrain on each date using last 10 dates). Correct?
---
### Author:  [@salty-francisco](https://forum.crunchdao.com/u/salty-francisco)
#### Post date: [August 21, 2023, 12:41pm UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/5 "2023-08-21T12:41:56Z")
yes, exactly!
---
### Author:  [@anxious-james](https://forum.crunchdao.com/u/anxious-james)
#### Post date: [August 25, 2023, 4:04pm UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/6 "2023-08-25T16:04:17Z")
> [@salty-francisco](#):
>
> np.log1p(target)
Can you explain the rationale behind this target transformation?
---
### Author:  [@tropical-lucas](https://forum.crunchdao.com/u/tropical-lucas)
#### Post date: [August 27, 2023, 5:57pm UTC](https://forum.crunchdao.com/t/my-approach-4-22-cv-3-704-public-lb/351/7 "2023-08-27T17:57:16Z")
I am curious now. How are the features selected ?