Amount Withdrawn Model Part 1

 

%load_ext autoreload

%autoreload 2

import sys
sys.path.insert(0, '../../Models/ALS')

from stark import *;hello()

/home/ec2-user/SageMaker/SUNDARESAN M S/Scripts/HandsOnML-CookBook
Data Science Environment Set!


from sklearn.model_selection import train_test_split
from sklearn.model_selection import cross_val_score,KFold
from sklearn.pipeline import make_pipeline

from sklearn.metrics import mean_absolute_error,mean_squared_error
from sklearn.metrics import make_scorer


from sklearn.linear_model import ElasticNet, Lasso
from sklearn.linear_model import LinearRegression
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor,  GradientBoostingRegressor
from xgboost import XGBRegressor
from lightgbm import LGBMRegressor

from matplotlib.pyplot import style
import matplotlib.pyplot as plt
import seaborn as sns
import pandas as pd

style.use('fivethirtyeight')

%matplotlib inline



dataset = pd.read_csv('../BSE DATASETS/ATM_Withdrawl_Prediction_Data.csv')

dataset.head()

	ID	Number of Shops Around ATM	ATM Zone	No of Other ATMs in 1 KM radius	Estimated Number of Houses in 1 KM Radius	ATM Placement	ATM TYPE	ATM Location TYPE	ATM looks	ATM Near	Avg Withdrawls Per Hour	ATM Attached to	Avg No of Withdrawls Per Week	Average Wait Time	Day Type	ATM RATING	ATM Prox	ATM Since	Holiday Sequence	AmountWithDrawn
0	TS000000001	66	RL	65	8450	Facing Road	Urban	Only WIthdraw	Normal	College	7	Building	856	3	Working	4	Attached to Branch	2008	WW	209500
1	TS000000002	26	RL	80	9600	Facing Road	Urban	Only WIthdraw	Normal	Temple	6	Building	1262	3	Working	3	Attached to Branch	2007	WW	184300
2	TS000000003	65	RL	68	11250	Facing Road	Town	Only WIthdraw	Normal	College	7	Building	920	3	Working	4	Attached to Branch	2008	WW	231500
3	TS000000004	80	RL	60	9550	Facing Road	Town	Only WIthdraw	Normal	Shopping Complex	7	Building	961	3	Working	4	2 ATM Machines (Same Bank)	2006	HH	143600
4	TS000000005	66	RL	84	14260	Facing Road	Town	Only WIthdraw	Normal	Park	8	Building	1145	4	Working	4	Attached to Branch	2008	WW	255600

exploration_dict= explore(dataset,['ID'])

Shape of the Dataset: (186880, 20)
head()

	ID	Number of Shops Around ATM	ATM Zone	No of Other ATMs in 1 KM radius	Estimated Number of Houses in 1 KM Radius	ATM Placement	ATM TYPE	ATM Location TYPE	ATM looks	ATM Near	Avg Withdrawls Per Hour	ATM Attached to	Avg No of Withdrawls Per Week	Average Wait Time	Day Type	ATM RATING	ATM Prox	ATM Since	Holiday Sequence	AmountWithDrawn
0	TS000000001	66	RL	65	8450	Facing Road	Urban	Only WIthdraw	Normal	College	7	Building	856	3	Working	4	Attached to Branch	2008	WW	209500
1	TS000000002	26	RL	80	9600	Facing Road	Urban	Only WIthdraw	Normal	Temple	6	Building	1262	3	Working	3	Attached to Branch	2007	WW	184300
2	TS000000003	65	RL	68	11250	Facing Road	Town	Only WIthdraw	Normal	College	7	Building	920	3	Working	4	Attached to Branch	2008	WW	231500
3	TS000000004	80	RL	60	9550	Facing Road	Town	Only WIthdraw	Normal	Shopping Complex	7	Building	961	3	Working	4	2 ATM Machines (Same Bank)	2006	HH	143600
4	TS000000005	66	RL	84	14260	Facing Road	Town	Only WIthdraw	Normal	Park	8	Building	1145	4	Working	4	Attached to Branch	2008	WW	255600

tail()

	ID	Number of Shops Around ATM	ATM Zone	No of Other ATMs in 1 KM radius	Estimated Number of Houses in 1 KM Radius	ATM Placement	ATM TYPE	ATM Location TYPE	ATM looks	ATM Near	Avg Withdrawls Per Hour	ATM Attached to	Avg No of Withdrawls Per Week	Average Wait Time	Day Type	ATM RATING	ATM Prox	ATM Since	Holiday Sequence	AmountWithDrawn
186875	TS000186876	62	RL	62	7917	Facing Road	Urban	Only WIthdraw	Normal	Lodge and Hotel	6	Building	953	3	Working	3	Attached to Branch	2007	WW	179700
186876	TS000186877	26	RL	85	13175	Facing Road	Urban	Only WIthdraw	Normal	Market	6	Building	2073	3	Working	3	Attached to Branch	2010	WW	215600
186877	TS000186878	80	RL	66	9042	Facing Road	Urban	Only WIthdraw	Normal	Shopping Complex	7	Building	1188	4	Working	4	Attached to Branch	2010	WW	267800
186878	TS000186879	21	RL	68	9717	Facing Road	Urban	Only WIthdraw	Normal	Highway	5	Petrol Bunk	1078	2	Working	4	Attached to Branch	2010	WW	147525
186879	TS000186880	21	RL	75	9937	Facing Road	Urban	Only WIthdraw	Normal	Church	5	Building	1256	3	Working	3	Attached to Branch	2008	WW	152100

Total number of columns: 20

array(['ID', 'Number of Shops Around ATM', 'ATM Zone',
       'No of Other ATMs in 1 KM radius',
       'Estimated Number of Houses in 1 KM Radius', 'ATM Placement',
       'ATM TYPE', 'ATM Location TYPE', 'ATM looks', 'ATM Near',
       'Avg Withdrawls Per Hour', 'ATM Attached to',
       'Avg No of Withdrawls Per Week', 'Average Wait Time', 'Day Type',
       'ATM RATING', 'ATM Prox', 'ATM Since', 'Holiday Sequence',
       'AmountWithDrawn'], dtype=object)

Missing Value Report:

	Missing_Value_Count	Missing_Value_Percentage
ID	0	0.0
Number of Shops Around ATM	0	0.0
Holiday Sequence	0	0.0
ATM Since	0	0.0
ATM Prox	0	0.0
ATM RATING	0	0.0
Day Type	0	0.0
Average Wait Time	0	0.0
Avg No of Withdrawls Per Week	0	0.0
ATM Attached to	0	0.0
Avg Withdrawls Per Hour	0	0.0
ATM Near	0	0.0
ATM looks	0	0.0
ATM Location TYPE	0	0.0
ATM TYPE	0	0.0
ATM Placement	0	0.0
Estimated Number of Houses in 1 KM Radius	0	0.0
No of Other ATMs in 1 KM radius	0	0.0
ATM Zone	0	0.0
AmountWithDrawn	0	0.0

Infering Features:

	Facing Road	Little Inside
ATM Placement	0.99589	0.00411

	Normal	New	Old and Dull
ATM looks	0.946575	0.044521	0.008904

	Urban	Town	Semi Urban	Bigger Towns
ATM TYPE	0.633562	0.331507	0.028082	0.006849

	Only WIthdraw	Deposit and Withdraw	Passbook Printing and Withdraw	Checkdrop and Withdraw
ATM Location TYPE	0.897945	0.043151	0.034247	0.024658

	Working	Festival	Strike	National Holiday
Day Type	0.953425	0.044521	0.00137	0.000685

	RL	RM	FV	RH	C
ATM Zone	0.788356	0.149315	0.044521	0.010959	0.006849

	WW	Long Weekend	HH	HW	WH
Holiday Sequence	0.828767	0.085616	0.069178	0.013699	0.00274

	Building	Petrol Bunk	Flat	Shop	House	Shed
ATM Attached to	0.781507	0.19589	0.008904	0.007534	0.004795	0.00137

	2+ ATMS Very Close	Attached to Branch	2 ATM Machines (Same Bank)	Other Bank ATM	Non Closed ATM	2 ATMS in same Building
ATM Prox	0.055479	0.59589	0.265068	0.060274	0.013014	0.010274

	count	mean	std	min	25%	50%	max	(-0.008, 4.0]	(4.0, 8.0]
Average Wait Time	186880.0	2.866438	0.815501	0.0	2.0	3.0	8.0	0.980137	0.019863

	count	mean	std	min	25%	75%	max	(1.997, 3.5]	(3.5, 5.0]
ATM RATING	186880.0	3.511644	0.663534	2.0	3.0	4.0	5.0	0.530137	0.469863

	count	mean	std	min	25%	50%	75%	max	(4.0, 7.0]	(7.0, 10.0]	(0.991, 4.0]
Avg Withdrawls Per Hour	186880.0	6.099315	1.382527	1.0	5.0	6.0	7.0	10.0	0.746575	0.156849	0.096575

	count	mean	std	min	25%	50%	75%	max	(1086.055, 22694.5]	(22694.5, 44089.0]	(44089.0, 65483.5]	(151061.5, 172456.0]	(193850.5, 215245.0]	(172456.0, 193850.5]
Estimated Number of Houses in 1 KM Radius	186880.0	10516.828082	9977.872801	1300.0	7553.5	9478.5	11601.5	215245.0	0.974658	0.016438	0.005479	0.00137	0.000685	0.0

	count	mean	std	min	25%	50%	75%	max	(769.8, 1205.6]	(1205.6, 1641.4]	(329.642, 769.8]	(1641.4, 2077.2]	(2077.2, 2513.0]	(2513.0, 2948.8]	(2948.8, 3384.6]	(4256.2, 4692.0]	(3820.4, 4256.2]
Avg No of Withdrawls Per Week	186880.0	1162.626712	386.456356	334.0	882.0	1087.0	1391.25	4692.0	0.503425	0.263699	0.113014	0.100685	0.014384	0.00274	0.00137	0.000685	0.0

	count	mean	std	min	25%	50%	75%	max	(2008.8, 2009.2]	(2006.8, 2007.2]	(2005.996, 2006.4]	(2007.6, 2008.0]	(2009.6, 2010.0]	(2009.2, 2009.6]
ATM Since	186880.0	2007.815753	1.327644	2006.0	2007.0	2008.0	2009.0	2010.0	0.231507	0.225342	0.215068	0.208219	0.119863	0.0

	count	mean	std	min	25%	50%	75%	max	(108350.0, 181300.0]	(181300.0, 254250.0]	(254250.0, 327200.0]	(34670.5, 108350.0]	(327200.0, 400150.0]	(400150.0, 473100.0]	(473100.0, 546050.0]	(546050.0, 619000.0]	(691950.0, 764900.0]	(619000.0, 691950.0]
AmountWithDrawn	186880.0	186128.713549	79466.216407	35400.0	135200.0	168050.0	218900.0	764900.0	0.489063	0.268076	0.097175	0.089822	0.035124	0.012778	0.003039	0.002697	0.00137	0.000856

	(20.821, 31.529]	(63.118, 73.647]	(52.588, 63.118]	(73.647, 84.176]	(84.176, 94.706]	(31.529, 42.059]	(115.765, 126.294]	(157.882, 168.412]	(42.059, 52.588]	(126.294, 136.824]	...	(178.941, 189.471]	(136.824, 147.353]	count	mean	std	min	25%	50%	75%	max
Number of Shops Around ATM	0.371816	0.15557	0.143044	0.054543	0.051375	0.043086	0.035766	0.034455	0.027702	0.023823	...	0.006202	0.0	186880.0	62.394542	42.389501	21.0	27.0	58.0	76.0	200.0

1 rows × 23 columns

	Highway	College	Residential	Church	Railway Station	Lodge and Hotel	Tourist Place	Shops	Market	Hospital	...	Upcoming Area	Food Joints	Resturant and Bar	Traffic Signal	IT park	Local Transport	Temple	Govt Buildings	2+ ATMS Very Close	Tasmac
ATM Near	0.15411	0.10274	0.077397	0.068493	0.058904	0.05411	0.05274	0.050685	0.05	0.040411	...	0.025342	0.019178	0.017123	0.013014	0.011644	0.010959	0.007534	0.006164	0.00411	0.00137

1 rows × 25 columns

	count	mean	std	min	25%	50%	75%	max	(51.207, 61.276]	(71.345, 81.414]	...	(31.069, 41.138]	(20.708, 31.069]	(101.552, 111.621]	(111.621, 121.69]	(121.69, 131.759]	(131.759, 141.828]	(151.897, 161.966]	(302.931, 313.0]	(161.966, 172.034]	(292.862, 302.931]
No of Other ATMs in 1 KM radius	186880.0	69.978082	22.60748	21.0	59.0	70.0	80.0	313.0	0.208219	0.202055	...	0.038356	0.032877	0.024658	0.012329	0.006849	0.00411	0.002055	0.00137	0.000685	0.0

1 rows × 24 columns

The following are the possible_date_features
  None

The following are the possible_categorical_features
  ATM Placement
  ATM looks
  ATM TYPE
  ATM Location TYPE
  Day Type
  ATM Zone
  Holiday Sequence
  ATM Attached to
  ATM Prox

The following are the possible_continuos_features
  Average Wait Time
  ATM RATING
  Avg Withdrawls Per Hour
  Estimated Number of Houses in 1 KM Radius
  Avg No of Withdrawls Per Week
  ATM Since
  AmountWithDrawn
  Number of Shops Around ATM
  No of Other ATMs in 1 KM radius

The following are the possible_categorical_binary
  ATM Placement

The following are the possible_categorical_multilabel
  ATM looks
  ATM TYPE
  ATM Location TYPE
  Day Type
  ATM Zone
  Holiday Sequence
  ATM Attached to
  ATM Prox

The following are the possible_categorical_string
  ATM Near

for key in list(exploration_dict.keys())[1:]:
    print(key)
    cols = exploration_dict[key]
    display(dataset[cols].head())

possible_categorical_features

	ATM Placement	ATM looks	ATM TYPE	ATM Location TYPE	Day Type	ATM Zone	Holiday Sequence	ATM Attached to	ATM Prox
0	Facing Road	Normal	Urban	Only WIthdraw	Working	RL	WW	Building	Attached to Branch
1	Facing Road	Normal	Urban	Only WIthdraw	Working	RL	WW	Building	Attached to Branch
2	Facing Road	Normal	Town	Only WIthdraw	Working	RL	WW	Building	Attached to Branch
3	Facing Road	Normal	Town	Only WIthdraw	Working	RL	HH	Building	2 ATM Machines (Same Bank)
4	Facing Road	Normal	Town	Only WIthdraw	Working	RL	WW	Building	Attached to Branch

possible_continuos_features

	Average Wait Time	ATM RATING	Avg Withdrawls Per Hour	Estimated Number of Houses in 1 KM Radius	Avg No of Withdrawls Per Week	ATM Since	AmountWithDrawn	Number of Shops Around ATM	No of Other ATMs in 1 KM radius
0	3	4	7	8450	856	2008	209500	66	65
1	3	3	6	9600	1262	2007	184300	26	80
2	3	4	7	11250	920	2008	231500	65	68
3	3	4	7	9550	961	2006	143600	80	60
4	4	4	8	14260	1145	2008	255600	66	84

possible_categorical_binary

	ATM Placement
0	Facing Road
1	Facing Road
2	Facing Road
3	Facing Road
4	Facing Road

possible_categorical_multilabel

	ATM looks	ATM TYPE	ATM Location TYPE	Day Type	ATM Zone	Holiday Sequence	ATM Attached to	ATM Prox
0	Normal	Urban	Only WIthdraw	Working	RL	WW	Building	Attached to Branch
1	Normal	Urban	Only WIthdraw	Working	RL	WW	Building	Attached to Branch
2	Normal	Town	Only WIthdraw	Working	RL	WW	Building	Attached to Branch
3	Normal	Town	Only WIthdraw	Working	RL	HH	Building	2 ATM Machines (Same Bank)
4	Normal	Town	Only WIthdraw	Working	RL	WW	Building	Attached to Branch

possible_categorical_string

	ATM Near
0	College
1	Temple
2	College
3	Shopping Complex
4	Park

possible_date_features

list(exploration_dict.keys())[1:]


    

['possible_categorical_features',
 'possible_continuos_features',
 'possible_categorical_binary',
 'possible_categorical_multilabel',
 'possible_categorical_string',
 'possible_date_features']

infer_schema(dataset)

	feature_name	data_type	number_of_classes	suggested_dtype
0	ID	object	186880	object
1	Number of Shops Around ATM	int64	120	int64
2	ATM Zone	object	5	object
3	No of Other ATMs in 1 KM radius	int64	110	int64
4	Estimated Number of Houses in 1 KM Radius	int64	1073	int64
5	ATM Placement	object	2	object
6	ATM TYPE	object	4	object
7	ATM Location TYPE	object	4	object
8	ATM looks	object	3	object
9	ATM Near	object	26	object
10	Avg Withdrawls Per Hour	int64	10	int64
11	ATM Attached to	object	6	object
12	Avg No of Withdrawls Per Week	int64	753	int64
13	Average Wait Time	int64	8	int64
14	Day Type	object	4	object
15	ATM RATING	int64	4	int64
16	ATM Prox	object	6	object
17	ATM Since	int64	5	int64
18	Holiday Sequence	object	5	object
19	AmountWithDrawn	int64	12929	int64

target =['AmountWithDrawn']
continuos_columns = [col for col in exploration_dict['possible_continuos_features'] if col not in target]

sns.boxplot(result_df.actuals)

<matplotlib.axes._subplots.AxesSubplot at 0x7f6cadb49748>

['possible_categorical_features',
 'possible_continuos_features',
 'possible_categorical_binary',
 'possible_categorical_multilabel',
 'possible_categorical_string',
 'possible_date_features']

dataset_cat = pd.get_dummies(dataset[exploration_dict['possible_categorical_features']])
dataset_continuos = dataset[continuos_columns]
dataset_fin = dataset_continuos.join(dataset_cat)

FSET = dataset_fin.columns

dataset_fin.shape,dataset[target[0]].shape

((186880, 47), (186880,))