Predictive Maintenance of Air Quality Data

# Load libraries
import pandas
from pandas.plotting import scatter_matrix
import matplotlib.pyplot as plt
from sklearn import model_selection
from sklearn.metrics import classification_report
from sklearn.metrics import confusion_matrix
from sklearn.metrics import accuracy_score
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.naive_bayes import GaussianNB
from sklearn.svm import SVC

# Load dataset
sensor_file = "./data/sensor_data.csv"
quality_file = "./data/quality_control_data.csv"
# names = ['weight', 'humidity', 'temperature', 'quality']
sensor_data = pandas.read_csv(sensor_file)
quality_data = pandas.read_csv(quality_file)

sensor_data.head(10)

	weight	humidity	temperature	prod_id
0	1030.871118	29.687881	71.995808	1
1	1044.961148	28.862453	68.468664	2
2	972.710479	37.951588	65.121344	3
3	1010.182509	25.076383	67.821336	4
4	970.039236	27.137886	72.931800	5
5	990.154359	32.422428	71.406207	6
6	965.660243	42.603619	65.876158	7
7	969.221212	31.655071	74.430054	8
8	976.495532	26.499721	69.866121	9
9	974.993517	38.644055	69.891709	10

quality_data.head(10)

	prod_id	quality
0	1	good
1	2	good
2	3	good
3	4	good
4	5	good
5	6	good
6	7	poor
7	8	good
8	9	good
9	10	good

rawdataset = sensor_data.merge(quality_data, on="prod_id")

rawdataset.head(5)

	weight	humidity	temperature	prod_id	quality
0	1030.871118	29.687881	71.995808	1	good
1	1044.961148	28.862453	68.468664	2	good
2	972.710479	37.951588	65.121344	3	good
3	1010.182509	25.076383	67.821336	4	good
4	970.039236	27.137886	72.931800	5	good

dataset = rawdataset.drop(columns='prod_id')
dataset.head(10)

	weight	humidity	temperature	quality
0	1030.871118	29.687881	71.995808	good
1	1044.961148	28.862453	68.468664	good
2	972.710479	37.951588	65.121344	good
3	1010.182509	25.076383	67.821336	good
4	970.039236	27.137886	72.931800	good
5	990.154359	32.422428	71.406207	good
6	965.660243	42.603619	65.876158	poor
7	969.221212	31.655071	74.430054	good
8	976.495532	26.499721	69.866121	good
9	974.993517	38.644055	69.891709	good

# shape
print(dataset.shape)

(3000, 4)

# descriptions
print(dataset.describe())

            weight     humidity  temperature
count  3000.000000  3000.000000  3000.000000
mean    999.940363    34.863581    69.962969
std      28.765904     5.755869     2.857898
min     950.017007    25.008023    65.000514
25%     975.552942    29.783650    67.522238
50%     998.875197    34.825848    69.890808
75%    1025.649219    39.887405    72.414522
max    1049.954013    44.986735    74.999312

# quality distribution
print(dataset.groupby('quality').size())

quality
good    2907
poor      93
dtype: int64

# box and whisker plots to show data distribution
dataset.plot(kind='box', subplots=True, layout=(2,2), sharex=False, sharey=False)
plt.show()

# check the histograms
dataset.hist()
plt.show()

# scatter plot matrix - anything useful here?
scatter_matrix(dataset)
plt.show()

# Split-out validation dataset
array = dataset.values
X = array[:,0:3]
Y = array[:,3]
validation_size = 0.20
seed = 8
X_train, X_validation, Y_train, Y_validation = model_selection.train_test_split(X, Y, test_size=validation_size, random_state=seed)

# Test options and evaluation metric
seed = 7
scoring = 'accuracy'

# Spot Check Algorithms
models = []
models.append(('LR', LogisticRegression(solver='lbfgs')))
models.append(('LDA', LinearDiscriminantAnalysis()))
models.append(('KNN', KNeighborsClassifier()))
models.append(('CART', DecisionTreeClassifier()))
models.append(('NB', GaussianNB()))
models.append(('SVM', SVC(gamma='auto')))
# evaluate each model in turn
results = []
names = []
for name, model in models:
    kfold = model_selection.KFold(n_splits=10, random_state=seed)
    cv_results = model_selection.cross_val_score(model, X_train, Y_train, cv=kfold, scoring=scoring)
    results.append(cv_results)
    names.append(name)
    msg = "%s: %f (%f)" % (name, cv_results.mean(), cv_results.std())
    print(msg)

LR: 0.976667 (0.008375)
LDA: 0.973750 (0.007229)
KNN: 0.992083 (0.005417)
CART: 0.998750 (0.002668)
NB: 0.994167 (0.003333)
SVM: 0.985417 (0.005966)

# Compare Algorithms
fig = plt.figure()
fig.suptitle('Comparison of ML Models')
ax = fig.add_subplot(111)
plt.boxplot(results)
ax.set_xticklabels(names)
plt.show()

# Make predictions on validation dataset
#knn = KNeighborsClassifier()
CART = DecisionTreeClassifier()
CART.fit(X_train, Y_train)
predictions = CART.predict(X_validation)
print(accuracy_score(Y_validation, predictions))
print(confusion_matrix(Y_validation, predictions))
print(classification_report(Y_validation, predictions))

0.9983333333333333
[[581   0]
 [  1  18]]
              precision    recall  f1-score   support

        good       1.00      1.00      1.00       581
        poor       1.00      0.95      0.97        19

    accuracy                           1.00       600
   macro avg       1.00      0.97      0.99       600
weighted avg       1.00      1.00      1.00       600

​

Now test some values of your own

testWeight = 1200
testHumidity = 60
testTemperature = 65
testPrediction = CART.predict([[testWeight,testHumidity,testTemperature]])
print(testPrediction)

['good']