├── README.md
├── data
    └── example.gif
├── main.py
└── requirements.txt


/README.md:
--------------------------------------------------------------------------------
 1 | # Interactive Web App with Streamlit and Scikit-learn
 2 | Explore different datasets and classifier. This tutorial should demonstrate how easy interactive web applications can be build with *streamlit*. Streamlit lets you create apps for your machine learning projects with simple Python scripts. See official [streamlit website](https://www.streamlit.io/) for more info.
 3 | 
 4 | ## Preview
 5 | ![Example of Streamlit|635x380](data/example.gif)
 6 | 
 7 | ## Watch the Tutorial
 8 |   [![Alt text](https://img.youtube.com/vi/Klqn--Mu2pE/hqdefault.jpg)](https://www.youtube.com/watch?v=Klqn--Mu2pE)
 9 | 
10 | ## Installation
11 | You need these dependencies:
12 | ```console
13 | pip install streamlit
14 | pip install scikit-learn
15 | pip install matplotlib
16 | ```
17 | 
18 | ## Usage
19 | Run
20 | ```console
21 | streamlit run main.py
22 | ```
23 | 


--------------------------------------------------------------------------------
/data/example.gif:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/patrickloeber/streamlit-demo/dd512ff9b8d6c407c02652053085749b39b70c66/data/example.gif


--------------------------------------------------------------------------------
/main.py:
--------------------------------------------------------------------------------
  1 | import streamlit as st 
  2 | import numpy as np 
  3 | 
  4 | import matplotlib.pyplot as plt
  5 | from sklearn import datasets
  6 | from sklearn.model_selection import train_test_split
  7 | 
  8 | from sklearn.decomposition import PCA
  9 | from sklearn.svm import SVC
 10 | from sklearn.neighbors import KNeighborsClassifier
 11 | from sklearn.ensemble import RandomForestClassifier
 12 | 
 13 | from sklearn.metrics import accuracy_score
 14 | 
 15 | st.title('Streamlit Example')
 16 | 
 17 | st.write("""
 18 | # Explore different classifier and datasets
 19 | Which one is the best?
 20 | """)
 21 | 
 22 | dataset_name = st.sidebar.selectbox(
 23 |     'Select Dataset',
 24 |     ('Iris', 'Breast Cancer', 'Wine')
 25 | )
 26 | 
 27 | st.write(f"## {dataset_name} Dataset")
 28 | 
 29 | classifier_name = st.sidebar.selectbox(
 30 |     'Select classifier',
 31 |     ('KNN', 'SVM', 'Random Forest')
 32 | )
 33 | 
 34 | def get_dataset(name):
 35 |     data = None
 36 |     if name == 'Iris':
 37 |         data = datasets.load_iris()
 38 |     elif name == 'Wine':
 39 |         data = datasets.load_wine()
 40 |     else:
 41 |         data = datasets.load_breast_cancer()
 42 |     X = data.data
 43 |     y = data.target
 44 |     return X, y
 45 | 
 46 | X, y = get_dataset(dataset_name)
 47 | st.write('Shape of dataset:', X.shape)
 48 | st.write('number of classes:', len(np.unique(y)))
 49 | 
 50 | def add_parameter_ui(clf_name):
 51 |     params = dict()
 52 |     if clf_name == 'SVM':
 53 |         C = st.sidebar.slider('C', 0.01, 10.0)
 54 |         params['C'] = C
 55 |     elif clf_name == 'KNN':
 56 |         K = st.sidebar.slider('K', 1, 15)
 57 |         params['K'] = K
 58 |     else:
 59 |         max_depth = st.sidebar.slider('max_depth', 2, 15)
 60 |         params['max_depth'] = max_depth
 61 |         n_estimators = st.sidebar.slider('n_estimators', 1, 100)
 62 |         params['n_estimators'] = n_estimators
 63 |     return params
 64 | 
 65 | params = add_parameter_ui(classifier_name)
 66 | 
 67 | def get_classifier(clf_name, params):
 68 |     clf = None
 69 |     if clf_name == 'SVM':
 70 |         clf = SVC(C=params['C'])
 71 |     elif clf_name == 'KNN':
 72 |         clf = KNeighborsClassifier(n_neighbors=params['K'])
 73 |     else:
 74 |         clf = clf = RandomForestClassifier(n_estimators=params['n_estimators'], 
 75 |             max_depth=params['max_depth'], random_state=1234)
 76 |     return clf
 77 | 
 78 | clf = get_classifier(classifier_name, params)
 79 | #### CLASSIFICATION ####
 80 | 
 81 | X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1234)
 82 | 
 83 | clf.fit(X_train, y_train)
 84 | y_pred = clf.predict(X_test)
 85 | 
 86 | acc = accuracy_score(y_test, y_pred)
 87 | 
 88 | st.write(f'Classifier = {classifier_name}')
 89 | st.write(f'Accuracy =', acc)
 90 | 
 91 | #### PLOT DATASET ####
 92 | # Project the data onto the 2 primary principal components
 93 | pca = PCA(2)
 94 | X_projected = pca.fit_transform(X)
 95 | 
 96 | x1 = X_projected[:, 0]
 97 | x2 = X_projected[:, 1]
 98 | 
 99 | fig = plt.figure()
100 | plt.scatter(x1, x2,
101 |         c=y, alpha=0.8,
102 |         cmap='viridis')
103 | 
104 | plt.xlabel('Principal Component 1')
105 | plt.ylabel('Principal Component 2')
106 | plt.colorbar()
107 | 
108 | #plt.show()
109 | st.pyplot(fig)
110 | 


--------------------------------------------------------------------------------
/requirements.txt:
--------------------------------------------------------------------------------
1 | streamlit
2 | matplotlib
3 | scikit-learn
4 | 


--------------------------------------------------------------------------------