first commit

LogReg.py

@@ -0,0 +1,78 @@
+import numpy as np
+import pandas as pd
+import matplotlib.pyplot as plt
+import seaborn as sns
+import streamlit as st
+from sklearn.preprocessing import StandardScaler
+
+
+class LogisticRegression():
+    def __init__(self, learning_rate =0.01, n_inputs=500):
+        self.learning_rate = learning_rate
+        self.n_inputs = n_inputs
+        self.coef_ = None
+        self.intercept_ = None
+        
+    def fit(self, X, y):
+        X = np.array(X)
+        n_samp, n_feat = X.shape
+        self.coef_ = np.zeros(n_feat)
+        self.intercept_ = 0
+        for _ in range(self.n_inputs):
+            lin_pred = [email protected]_ + self.intercept_
+            predictions = 1/(1+np.exp(-lin_pred))
+            
+            di = predictions-y
+            dc = X * di.reshape(-1, 1)
+            
+            self.coef_ -= self.learning_rate * dc.mean(axis=0)
+            self.intercept_ -= self.learning_rate * di.mean()
+
+st.title('Выберите файл')
+uploaded_file = st.file_uploader(label='', label_visibility='collapsed', type=["csv"])
+if uploaded_file:
+    df = pd.read_csv(uploaded_file).drop('Unnamed: 0', axis=1)
+    st.write(df.head(3))
+    target = st.selectbox('Выберите столбец, который будет целью обучения', *[df.columns])
+    features = st.multiselect('Выберите признаки', *[df.columns])
+    if features:
+        new_df = df.copy()
+        ss = StandardScaler()
+        new_df[features] = ss.fit_transform(new_df[features])
+
+        logreg = LogisticRegression(0.01, 5000)
+        logreg.fit(new_df[features], new_df[target].to_numpy())
+
+        outdict = pd.DataFrame(logreg.coef_.reshape(1,-1), columns=features)
+        st. write(outdict)
+
+        st.write('# Графики :chart_with_upwards_trend:')
+        x = st.selectbox("Выберите X", *[df.columns])
+        y = st.selectbox('Выберите Y', *[df.columns])
+        if x and y:
+            choose = st.selectbox('Выберите график', ['Диаграмма рассеяния', 'Столбчатая диаграмма', 'Линейная диаграмма'])
+            if choose == 'Диаграмма рассеяния':
+                plt.figure(figsize=(20, 15))
+                plt.scatter(x=new_df[x], y=new_df[y])
+                plt.xlabel(x)
+                plt.ylabel(y)
+                plt.grid(True)
+                st.pyplot(plt)
+            elif choose == 'Столбчатая диаграмма':
+                plt.figure(figsize=(20, 15))
+                plt.bar(x=new_df[x].sort_values(), height=new_df[y].sort_values())
+                plt.xlabel(x)
+                plt.ylabel(y)
+                plt.grid(True)
+                st.pyplot(plt)
+            elif choose == 'Линейная диаграмма':
+                plt.figure(figsize=(20, 15))
+                plt.plot(new_df[x].sort_values(), new_df[y].sort_values())
+                plt.xlabel(x)
+                plt.ylabel(y)
+                plt.grid(True)
+                st.pyplot(plt)
+
+
+
+