ml_app.py

import streamlit as st
import joblib
import os
import numpy as np


attrib_info = """
#### Attribute Information:
    - Age 1.20-65
    - Sex 1. Male, 2.Female
    - Polyuria 1.Yes, 2.No.
    - Polydipsia 1.Yes, 2.No.
    - sudden weight loss 1.Yes, 2.No.
    - weakness 1.Yes, 2.No.
    - Polyphagia 1.Yes, 2.No.
    - Genital thrush 1.Yes, 2.No.
    - visual blurring 1.Yes, 2.No.
    - Itching 1.Yes, 2.No.
    - Irritability 1.Yes, 2.No.
    - delayed healing 1.Yes, 2.No.
    - partial paresis 1.Yes, 2.No.
    - muscle stiness 1.Yes, 2.No.
    - Alopecia 1.Yes, 2.No.
    - Obesity 1.Yes, 2.No.
    - Class 1.Positive, 2.Negative.
"""

label_dict = {"No":0,"Yes":1}
gender_map={"Female":0, "Male":1}
target_label_map = {"Negative":0,"Positive":1}

def get_fvalue(val):
    feature_dict={"No":0, "Yes":1}
    for key, value in feature_dict.items():
        if val==key:
            return value

def get_value(val, my_dict):
    for key, value in my_dict.items():
        if val==key:
            return value



@st.cache_data
def load_model(model_file):
    loaded_model=joblib.load(open(os.path.join(model_file), "rb"))
    return loaded_model



def run_ml_app():
    st.subheader("From ML Prediction")

    with st.expander("Attribute Info"):
        st.markdown(attrib_info)    

    col1, col2= st.columns(2)

    with col1:
        age = st.number_input("Age",10,100)
        gender = st.radio("Gender",("Female","Male"))
        polyuria = st.radio("Polyuria",["No","Yes"])
        polydipsia = st.radio("Polydipsia",["No","Yes"]) 
        sudden_weight_loss = st.selectbox("Sudden_weight_loss",["No","Yes"])
        weakness = st.radio("weakness",["No","Yes"]) 
        polyphagia = st.radio("polyphagia",["No","Yes"]) 
        genital_thrush = st.selectbox("Genital_thrush",["No","Yes"]) 

    with col2:
        visual_blurring = st.selectbox("Visual_blurring",["No","Yes"])
        itching = st.radio("itching",["No","Yes"]) 
        irritability = st.radio("irritability",["No","Yes"]) 
        delayed_healing = st.radio("delayed_healing",["No","Yes"]) 
        partial_paresis = st.selectbox("Partial_paresis",["No","Yes"])
        muscle_stiffness = st.radio("muscle_stiffness",["No","Yes"]) 
        alopecia = st.radio("alopecia",["No","Yes"]) 
        obesity = st.select_slider("obesity",["No","Yes"])

    with st.expander("Your Selected Options"):
        result={"Age":age, 
                "gender":gender, 
                "polyuria":polyuria,
                "polydipsia":polydipsia,
                "sudden_weight_loss":sudden_weight_loss,
                "weakness":weakness,
                "polyphagia":polyphagia,
                "genital_thrush":genital_thrush,
                "visual_blurring":visual_blurring, 
                "itching":itching, 
                "irritability": irritability,
                "delayed_healing":delayed_healing, 
                "partial_paresis":partial_paresis,
                "muscle_stiffness":muscle_stiffness,
                "alopecia":alopecia,
                "obesity":obesity}
        st.write("In JSON format:")
        st.write(result)

        encoded_result=[]
        for i in result.values():
            if type(i)==int:
                encoded_result.append(i)
            elif i in ["Female","Male"]:
                res=get_value(i,gender_map)
                encoded_result.append(res)
            else:
                encoded_result.append(get_fvalue(i))
        st.write("In LIST format")
        st.write(encoded_result)

    with st.expander("Prediction Results"):
        single_sample=np.array(encoded_result).reshape(1,-1)
        st.write(single_sample)

        model=load_model("models/logistic_regression_model_diabetes_21_oct_2020.pkl")
        prediction=model.predict(single_sample)
        pred_prob=model.predict_proba(single_sample)
        st.write(prediction)
        st.write(pred_prob)

        if prediction==1:
            st.warning("Positive Risk {}".format(prediction[0]))
            pred_probability_score={"Negative DM Risk":round(pred_prob[0][0]*100,2), "Positive DM Risk":round(pred_prob[0][1]*100,2)}
            st.write(pred_probability_score)
        else:
            st.success("Negative Risk {}".format(prediction[0]))
            pred_probability_score={"Negative DM Risk":round(pred_prob[0][0]*100,2), "Positive DM Risk":round(pred_prob[0][1]*100,2)}
            st.write(pred_probability_score)