hotel booking predictor

in this i tried to find the best model for a imbalanced dataset with 36 million rows. for this dataset we are gonna do some classification. unfortunately link of this dataset is not accessible at the time of writing this. main challenges of this code was:

• data was imbalanced
• data wasnt representative of what we are asked for

for preprocessing of this code i did lots of visualizations and based on some of them i removed what seemed to be outliers according to its z-score or IQR. after removing outliers i tried to tackle issues caused by being a imbalanced dataset. i tried different solutions and tried to find a good threshold for our predictions. with our final dataset we going after checking different models to see which will perform the best.

lets look at dataset:

	user	search_date	channel	is_mobile	is_package	destination	checkIn_date	checkOut_date	n_adults	n_children	n_rooms	hotel_category	is_booking
0	u461899	2019-01-07 00:00:02	c9	False	False	d669	2019-03-14	2019-03-15	2	1	1	g41	False
1	u13796	2019-01-07 00:00:06	c9	False	False	d8821	2019-01-19	2019-01-26	1	0	1	g58	False
2	u1128575	2019-01-07 00:00:06	c9	False	False	d25064	2019-01-19	2019-01-22	1	0	1	g91	False
3	u1080476	2019-01-07 00:00:09	c9	False	True	d7635	2019-05-29	2019-06-05	2	0	1	g10	False
4	u1080476	2019-01-07 00:00:17	c9	False	True	d7635	2019-05-29	2019-06-05	2	0	1	g10	False
...	...	...	...	...	...	...	...	...	...	...	...	...	...
34742970	u553256	2020-11-30 23:59:48	c2	True	True	d45532	2020-12-07	2020-12-08	2	0	1	g48	False
34742971	u529472	2020-11-30 23:59:49	c9	False	False	d8279	2020-12-27	2021-01-02	2	2	1	g18	False
34742972	u18236	2020-11-30 23:59:53	c4	False	False	d20275	2021-04-22	2021-04-25	1	0	1	g5	False
34742973	u10888	2020-11-30 23:59:54	c9	False	False	d19371	2020-12-29	2020-12-30	2	0	1	g17	False
34742974	u233344	2020-11-30 23:59:55	c9	False	False	d22862	2021-08-16	2021-08-18	2	0	1	g44	False

34742975 rows × 13 columns