Source: View original notebook on GitHub
Category: Machine Learning / Learn ML
Challenge - Air Pollution(https://www.kaggle.com/c/air-pollution-prediction/overview)
It is winter time in Delhi so Cody decided to go for a walk to the news stand and on reaching was surprised to see the air quality index of Delhi in the newspaper . So he decided to collect the air samples of different locations and then took these samples to his lab where he extracted five features of the air he collected which can be used to predict the air quality index and combined it with the air quality index given in the newspapers. You are provided with the data collected by Cody and your job is to design a machine learning model , which is given by the features extracted by Cody to predict air quality.
submit file like sample_submittion format :-
| Id | target |
|---|---|
| 0 | ans0 |
| 1 | ans1 |
| 2 | ans2 |
| 3 | ans3 |
| 4 | ans4 |
| 5 | ans5 |
| 6 | ans6 |
# loading Dataset
import pandas as pd
train = pd.read_csv('Datasets/Assignment2_Train.csv')
train.head()
Output:
feature_1 feature_2 feature_3 feature_4 feature_5 target
0 0.293416 -0.945599 -0.421105 0.406816 0.525662 -82.154667
1 -0.836084 -0.189228 -0.776403 -1.053831 0.597997 -48.897960
2 0.236425 0.132836 -0.147723 0.699854 -0.187364 77.270371
3 0.175312 0.143194 -0.581111 -0.122107 -1.292168 -2.988581
4 -1.693011 0.542712 -2.798729 -0.686723 1.244077 -37.596722
train.shape # 5 are features; 1 is target
Output:
(1600, 6)
train.columns
Output:
Index(['feature_1', 'feature_2', 'feature_3', 'feature_4', 'feature_5',
'target'],
dtype='object')
X = train[train.columns[:-1]]
X.head()
Output:
feature_1 feature_2 feature_3 feature_4 feature_5
0 0.293416 -0.945599 -0.421105 0.406816 0.525662
1 -0.836084 -0.189228 -0.776403 -1.053831 0.597997
2 0.236425 0.132836 -0.147723 0.699854 -0.187364
3 0.175312 0.143194 -0.581111 -0.122107 -1.292168
4 -1.693011 0.542712 -2.798729 -0.686723 1.244077
Y = train[train.columns[-1]]
Y.head()
Output:
0 -82.154667
1 -48.897960
2 77.270371
3 -2.988581
4 -37.596722
Name: target, dtype: float64
X.shape, Y.shape
Output:
((1600, 5), (1600,))
type(X)
Output:
pandas.core.frame.DataFrame
import pandas as pd
# preprocessing
from sklearn.preprocessing import StandardScaler
s = StandardScaler()
X = pd.DataFrame(s.fit_transform(X))
# output for test file
test = pd.read_csv('Datasets/Assignment2_Test.csv')
test = s.transform(test.values)
type(X)
Output:
pandas.core.frame.DataFrame
X.describe()
Output:
0 1 2 3 4
count 1.600000e+03 1.600000e+03 1.600000e+03 1.600000e+03 1.600000e+03
mean -2.331468e-17 -1.776357e-17 -1.776357e-17 2.331468e-17 8.881784e-18
std 1.000313e+00 1.000313e+00 1.000313e+00 1.000313e+00 1.000313e+00
min -3.394334e+00 -3.218189e+00 -3.073464e+00 -3.154539e+00 -2.927091e+00
25% -6.532217e-01 -6.631960e-01 -6.544315e-01 -6.560276e-01 -6.417809e-01
50% -4.487509e-03 -1.582564e-02 3.151454e-03 1.244233e-02 -2.609701e-02
75% 6.800261e-01 6.589081e-01 6.758504e-01 6.772709e-01 6.522049e-01
max 3.292885e+00 3.393682e+00 3.223719e+00 2.977582e+00 3.383015e+00
Y.describe()
Output:
count 1600.000000
mean 0.318835
std 110.741562
min -379.829794
25% -71.897040
50% -0.610665
75% 71.226603
max 337.643014
Name: target, dtype: float64
from sklearn.model_selection import train_test_split
X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.25,random_state=101)
X_train.shape, X_test.shape, Y_train.shape, Y_test.shape
Output:
((1200, 5), (400, 5), (1200,), (400,))
from sklearn.linear_model import LinearRegression
model = LinearRegression()
model.fit(X_train, Y_train)
Output:
LinearRegression(copy_X=True, fit_intercept=True, n_jobs=None,
normalize=False)
model.coef_
Output:
array([29.85207419, 94.83165412, 8.05996353, 45.23534964, 2.34253763])
model.intercept_
Output:
0.6828984772412013
Y_pred = model.predict(X_test)
model.score(X_test,Y_test)
Output:
0.9682476621478511
from sklearn.metrics import r2_score
r2_score(Y_test,Y_pred)
Output:
0.9682476621478511
%matplotlib inline
import matplotlib.pyplot as plt
plt.scatter(Y_test,Y_pred) # approximately linear
Output:
<matplotlib.collections.PathCollection at 0x1286ca30>
# output for test file
test = pd.DataFrame(test)
test.head()
Output:
0 1 2 3 4
0 1.014156 2.059621 -0.219462 -2.336264 -1.014474
1 -0.380266 0.960186 0.663580 0.734423 -0.375628
2 -1.031635 0.969787 -0.359367 -1.788688 0.453005
3 -2.501395 0.560382 0.638368 -0.555246 -1.355645
4 -0.389949 -0.789089 1.387127 -0.497137 0.408771
test.describe()
Output:
0 1 2 3 4
count 400.000000 400.000000 400.000000 400.000000 400.000000
mean -0.025996 0.014487 0.033092 0.007419 0.019432
std 0.982084 1.000121 1.030062 1.017343 0.958216
min -2.576064 -2.902958 -2.692526 -2.949381 -2.560900
25% -0.791048 -0.600358 -0.727477 -0.676770 -0.614510
50% -0.008899 -0.048541 -0.064106 0.001629 0.036103
75% 0.679685 0.749801 0.789482 0.687178 0.587112
max 3.250423 2.564306 2.920374 2.659714 2.921637
output = model.predict(test)
output.shape
Output:
(400,)
type(output)
Output:
numpy.ndarray
output.dtype
Output:
dtype('float64')
df = pd.DataFrame(output, index = [i for i in range(output.shape[0])], columns= ['target'])
df.head()
Output:
target
0 116.447831
1 118.077583
2 -20.894286
3 -43.994147
4 -96.138932
df.index.name = 'Id'
df.head()
Output:
target
Id
0 116.447831
1 118.077583
2 -20.894286
3 -43.994147
4 -96.138932
df.to_csv('Datasets/Assignment2_output.csv') # 33 rank with score of 0.96802
Using code from Scratch
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
# theta will be np.array([theta0,theta1,.............,thetaN])
def error(X,Y,theta):
hx = hypothesis(X,theta)
return (0.5) * np.sum((hx-Y)**2)
def hypothesis(X,theta):
# returning h(Xi) for each sample (Xi) hence returning an array of size X.shape[0]
data = np.sum(X*theta[1:], axis=1) + theta[0]
return data
def gradient(X,Y,theta):
grad = np.zeros(X.shape[1]+1) # n-featured data have (n+1) parameters.
hx = hypothesis(X,theta)
grad[0] = np.sum((hx-Y))
for i in range(1,len(grad)) :
# d(cost) / d(theta[i]) = sum((hx-y)*x[i])
grad[i] = np.sum( (hx-Y) * X[:,i-1])
return grad
def gradientDescent_multivariate(X,Y,learning_rate=0.0001):
theta = np.zeros(X.shape[1]+1)
error_list = []
err = error(X,Y,theta)
error_list.append(err)
while True:
grad = gradient(X,Y,theta)
theta = theta - learning_rate * grad
err = error(X,Y,theta)
error_change = abs(err - error_list[-1])
error_list.append(err)
if error_change < 0.00001:
break
return theta,error_list
theta,error = gradientDescent_multivariate(X_train.values,Y_train.values)
theta
Output:
array([ 0.68287854, 29.85199475, 94.83151151, 8.05984701, 45.23524455,
2.34249223])
# prediction
test.shape
Output:
(400, 5)
hypo = theta[0] + theta[1]*test[0] + theta[2]*test[1] + theta[3]*test[2] + theta[4]*test[3] + theta[5]*test[4]
hypo.shape
Output:
(400,)
output = hypo.values
type(output)
Output:
numpy.ndarray
output.dtype
Output:
dtype('float64')
df = pd.DataFrame(output, index = [i for i in range(output.shape[0])], columns= ['target'])
df.head()
Output:
target
0 116.447754
1 118.077318
2 -20.894153
3 -43.994003
4 -96.138936
df.index.name = 'Id'
df.head()
Output:
target
Id
0 116.447754
1 118.077318
2 -20.894153
3 -43.994003
4 -96.138936
df.to_csv('Datasets/Assignment2_output_scratch.csv') # 33 rank with score of 0.96802