Baseline Submission for DBSRA

[aicrowd-bot]

Contribute Download Execute In Colab

Baseline submission for the challenge DBSRA¶

Author - Pulkit Gera¶

In [ ]:

!pip install numpy
!pip install pandas
!pip install sklearn

In [1]:

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split 
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import LabelEncoder
from sklearn import metrics

Download data¶

The first step is to download out train test data. We will be training a classifier on the train data and make predictions on test data. We submit our predictions

In [ ]:

!wget https://s3.eu-central-1.wasabisys.com/aicrowd-public-datasets/aicrowd_educational_dbsra/data/public/test.csv
!wget https://s3.eu-central-1.wasabisys.com/aicrowd-public-datasets/aicrowd_educational_dbsra/data/public/train.csv

Load Data¶

We use pandas library to load our data. Pandas loads them into dataframes which helps us analyze our data easily. Learn more about it here

In [2]:

train_data = pd.read_csv('train.csv')

Clean and Analyse Data¶

In [3]:

train_data = train_data.drop('encounter_id',1)
train_data = train_data.drop('patient_nbr',1)
train_data.head()

Out[3]:

	race	gender	age	weight	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	payer_code	medical_specialty	...	citoglipton	insulin	glyburide-metformin	glipizide-metformin	glimepiride-pioglitazone	metformin-rosiglitazone	metformin-pioglitazone	change	diabetesMed	readmitted
0	AfricanAmerican	Female	[70-80)	?	1	1	7	2	?	InternalMedicine	...	No	Steady	No	No	No	No	No	No	Yes	1
1	Caucasian	Female	[90-100)	?	3	1	1	8	SP	Pulmonology	...	No	Down	No	No	No	No	No	Ch	Yes	1
2	Caucasian	Female	[80-90)	?	1	2	7	1	MC	Osteopath	...	No	Steady	No	No	No	No	No	No	Yes	0
3	Caucasian	Male	[60-70)	?	3	1	6	6	MC	Radiologist	...	No	Steady	No	No	No	No	No	Ch	Yes	0
4	?	Female	[70-80)	?	1	3	6	3	UN	InternalMedicine	...	No	No	No	No	No	No	No	No	No	0

5 rows × 48 columns

Since most of the columns have categorical columns we have to convert it into integers. The most basic way is to do an Ordinal Mapping. Note: Here we have not replaced question marks with some other data and they are also accounted into ordinal mapping.

In [5]:

labelencoder = LabelEncoder()
n_train_data = train_data
for col in train_data.columns:
    s = train_data[col]
    if s.dtype == 'O':
        s = labelencoder.fit_transform(s)
        n_train_data[col] = s
n_train_data.head()

Out[5]:

	race	gender	age	weight	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	payer_code	medical_specialty	...	citoglipton	insulin	glyburide-metformin	glipizide-metformin	glimepiride-pioglitazone	metformin-rosiglitazone	metformin-pioglitazone	change	diabetesMed	readmitted
0	1	0	7	1	1	1	7	2	0	19	...	0	2	1	0	0	0	0	1	1	1
1	3	0	9	1	3	1	1	8	15	51	...	0	0	1	0	0	0	0	0	1	1
2	3	0	8	1	1	2	7	1	8	30	...	0	2	1	0	0	0	0	1	1	0
3	3	1	6	1	3	1	6	6	8	52	...	0	2	1	0	0	0	0	0	1	0
4	0	0	7	1	1	3	6	3	16	19	...	0	1	1	0	0	0	0	1	0	0

5 rows × 48 columns

Split Data into Train and Validation¶

Now we want to see how well our classifier is performing, but we dont have the test data labels with us to check. What do we do ? So we split our dataset into train and validation. The idea is that we test our classifier on validation set in order to get an idea of how well our classifier works. This way we can also ensure that we dont overfit on the train dataset. There are many ways to do validation like k-fold,leave one out, etc

In [6]:

X = n_train_data.drop('readmitted',1)
y = n_train_data['readmitted']
# Validation testing
X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)

Here we have selected the size of the validation data to be 20% of the total data. You can change it and see what effect it has on the accuracies. To learn more about the train_test_split function click here.

Define the Classifier and Train¶

Now we come to the juicy part. We have fixed our data and now we train a classifier. The classifier will learn the function by looking at the inputs and corresponding outputs. There are a ton of classifiers to choose from some being Logistic Regression, SVM, Random Forests, Decision Trees, etc.
Tip: A good model doesnt depend solely on the classifier but on the features(columns) you choose. So make sure to play with your data and keep only whats important.

In [7]:

classifier = LogisticRegression()
classifier.fit(X_train,y_train)

# from sklearn.svm import SVC
# clf = SVC(gamma='auto')
# clf.fit(X_train, y_train)

# from sklearn import tree
# clf = tree.DecisionTreeClassifier()
# clf = clf.fit(X_train, y_train)

/home/gera/anaconda3/lib/python3.6/site-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.
  FutureWarning)
/home/gera/anaconda3/lib/python3.6/site-packages/sklearn/linear_model/logistic.py:469: FutureWarning: Default multi_class will be changed to 'auto' in 0.22. Specify the multi_class option to silence this warning.
  "this warning.", FutureWarning)

Out[7]:

LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
                   intercept_scaling=1, l1_ratio=None, max_iter=100,
                   multi_class='warn', n_jobs=None, penalty='l2',
                   random_state=None, solver='warn', tol=0.0001, verbose=0,
                   warm_start=False)

We have used Logistic Regression as a classifier here and set few of the parameteres. But one can set more parameters and increase the performance. To see the list of parameters visit here.
Also given are SVM and Decision Tree examples. Check out SVM's parameters here and Decision Tree's here

got a warning! Dont worry, its just beacuse the number of iteration is very less(defined in the classifier in the above cell).Increase the number of iterations and see if the warning vanishes.Do remember increasing iterations also increases the running time.( Hint: max_iter=500)

Predict on Validation¶

Now we predict our trained classifier on the validation set and evaluate our model

In [8]:

y_pred = classifier.predict(X_val)

In [9]:

df = pd.DataFrame({'Actual': y_val, 'Predicted': y_pred})
df1 = df.head(25)
df1

Out[9]:

	Actual	Predicted
26342	1	0
59142	1	0
57537	1	0
58128	0	0
29821	1	0
62897	0	0
43572	0	0
62329	2	0
44309	0	0
20882	0	0
49075	0	0
20668	0	0
76856	1	0
32858	1	1
74292	1	0
80549	1	0
8588	1	0
57768	1	0
10658	1	0
51569	0	0
59914	1	0
32874	0	0
54656	1	0
77456	0	0
35300	0	0

Evaluate the Performance¶

We use the same metrics as that will be used for the test set.
F1 score and Log Loss are the metrics for this challenge

In [10]:

print('F1 score Score:', metrics.f1_score(y_val, y_pred,average='micro'))

F1 score Score: 0.5688110513843131

Load Test Set¶

Load the test data now

In [11]:

test_data = pd.read_csv('test.csv')

In [12]:

test_data = test_data.drop('encounter_id',1)
test_data = test_data.drop('patient_nbr',1)
n_test_data = test_data
for col in test_data.columns:
    s = test_data[col]
    if s.dtype == 'O':
        s = labelencoder.fit_transform(s)
        n_test_data[col] = s
n_test_data.head()

Out[12]:

	race	gender	age	weight	admission_type_id	discharge_disposition_id	admission_source_id	time_in_hospital	payer_code	medical_specialty	...	examide	citoglipton	insulin	glyburide-metformin	glipizide-metformin	glimepiride-pioglitazone	metformin-rosiglitazone	metformin-pioglitazone	change	diabetesMed
0	3	0	7	1	1	1	6	11	15	16	...	0	0	2	1	0	0	0	0	1	1
1	3	1	5	1	1	1	1	1	6	0	...	0	0	1	1	0	0	0	0	1	1
2	3	0	6	1	3	6	1	4	6	0	...	0	0	1	1	0	0	0	0	1	1
3	3	1	3	1	2	1	1	12	4	10	...	0	0	1	1	0	0	0	0	1	1
4	1	0	6	1	1	2	7	1	0	0	...	0	0	1	1	0	0	0	0	1	1

5 rows × 47 columns

Predict Test Set¶

Time for the moment of truth! Predict on test set and time to make the submission.

In [13]:

y_test = classifier.predict(test_data)

In [14]:

df = pd.DataFrame(y_test,columns=['readmitted'])
df.to_csv('submission.csv',index=False)

To download the generated csv in collab run the below command¶

In [ ]:

from google.colab import files
files.download('submission.csv')

To participate in the challenge click here

In [ ]: