app.py

import uvicorn
import base64
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px
import random
from statsmodels.stats.power import TTestIndPower
from collections import Counter
from fastapi import FastAPI
from simulation import simulation
#from power_analysis import power_analysis
from fastapi.responses import StreamingResponse
from fastapi import FastAPI, UploadFile,File
from pydantic import BaseModel
import io 
#from flask import Flask, request
app = FastAPI()


@app.get('/')
def index():
    return {'message': 'Hello, World'}

@app.post('/simulate_dataset')
def dataset_simulation(data:simulation):
  
  """
  This function simulates a dataset given a set of population parameters, description of population parameters are as given below.


  User Defined Inputs:
  1.size=Population Sample size
  2.percentage_alcoholism/depression/tobacco=Percentage of the the respective condition in the population
  3.percentage_alcoholism_depression(and other dual variable overlaps)= Percentage of the overlap of respective two condtions in the populations
  4.percentage_tobacco_alcoholism_depression= Percentage of the overlap of all 3 cases in the population
  5.age, gender, bmi, edu: A list consisting of the percentages of distribution in the various categorical buckets for the following variables:
  eg; age=[18,60,35,15]:min,max,mean,std dev 
  age:4 elements; gender: 2 buckets; bmi: 3 buckets; edu: 4 buckets
  6.Treatment_noth/treatment_1_conditions=The percentage of good treatment outcomes for the population with the following number of conditions
  """
  print("Type is",type(data.dict))
  try:
    data= data.dict()
    size = data['size']
    percentage_alc_only = data['percentage_alc_only']
    percentage_dep_only = data['percentage_dep_only'] 
    percentage_tobacco_only = data['percentage_tobacco_only']
    percentage_alc_dep = data['percentage_alc_dep']
    percentage_alc_tobacco = data['percentage_alc_tobacco']
    percentage_dep_tobacco = data['percentage_dep_tobacco'] 
    percentage_tobacco_alcoholism_depression = data['percentage_tobacco_alcoholism_depression']
    treatment_noth = data['treatment_noth']
    treatment_1_conditions = data['treatment_1_conditions']
    treatment_2_conditions = data['treatment_2_conditions']
    treatment_3_conditions = data['treatment_3_conditions']
    gender = data['gender']
    bmi = data['bmi']
    edu = data['edu']
    seed = data['seed']
    age = data['age']

    zeros=1-(percentage_alc_only + percentage_dep_only  + percentage_tobacco_only + percentage_alc_dep + percentage_alc_tobacco + percentage_dep_tobacco + percentage_tobacco_alcoholism_depression)
    
    print("zeros:{}".format(zeros))
    np.random.seed(2020)
    condition=np.random.choice(8,size,p=[zeros, percentage_alc_only, percentage_dep_only,percentage_alc_dep,percentage_tobacco_only,percentage_alc_tobacco, percentage_dep_tobacco, percentage_tobacco_alcoholism_depression])
    alcoholism=np.where((condition == 1) | (condition==3) | (condition==5) | (condition==7),1,0)
    depression=np.where((condition == 2) | (condition==3) | (condition==6) | (condition==7),1,0)
    
    alc_only=np.where(condition == 1,1,0)
    dep_only=np.where(condition == 2,1,0)
    tobacco_only=np.where(condition == 4,1,0)

    alc_dep=np.where(condition == 3,1,0)
    alc_tobacco=np.where(condition == 5,1,0)
    dep_tobacco=np.where(condition == 6,1,0)

    tobacco=np.where((condition == 4) | (condition==5) | (condition==6) | (condition==7),1,0)
    all_three=np.where(condition == 7,1,0)
    age_1= np.random.normal(loc=age[2], scale=age[3], size=size)
    age_1=np.clip(age_1, a_min=age[0], a_max=age[1])
    age_1 = np.round(age_1).astype(int)
    sex=np.random.choice(2,size,p=[gender[0],gender[1]])
    body_mass=np.random.choice(3,size,p=[bmi[0],bmi[1],bmi[2]])
    education=np.random.choice(4,size,p=[edu[0],edu[1],edu[2],edu[3]])
    cavitation = np.random.choice(2,size,p=[0.5,0.5])
    ttd = np.random.normal(loc=7, scale=3, size=size)
    df = pd.DataFrame(
      {
          'idx': np.arange(1, size+1),
          'age': age_1,
          'gender': sex,
          'bmi': body_mass,
          'education': education,
          'cavitation': cavitation,
          'TTD': ttd,
          'alcoholism': alcoholism,
          'depression': depression,
          'tobacco': tobacco,
          'alcohol_only':alc_only,
          'depression_only':dep_only,
          'tobacco_only': tobacco_only,
          'alcoholism+depression':alc_dep,
          'alcoholism+tobacco':alc_tobacco,
          'depression+tobacco':dep_tobacco,
          'tobacco+alcohol+smoking':all_three
      }
    )
    intervention_arr=[]
    choices_alc_only=['NAlc','A']
    choices_dep_only=['ND','D']
    choices_tobacco_only=['NT','T']
    choices_alc_dep_only=['NAD','AD']
    choices_alc_tobacco_only=['NAT','AT']
    choices_dep_tobacco_only=['NDT','DT']
    choices_all_3=['NADT','ADT']
    random.seed(seed)
    weights=[0.5,0.5]
    for i in range(size):
      if(df['alcohol_only'][i]==1):
        #intervention_arr.append(np.random.binomial(1,0.5,size=1))
        intervention_arr.append(random.choices(choices_alc_only,weights=weights)[0])
      if(df['depression_only'][i]==1):
        intervention_arr.append(random.choices(choices_dep_only,weights=weights)[0])
      if(df['tobacco_only'][i]==1):
        intervention_arr.append(random.choices(choices_tobacco_only,weights=weights)[0])
      if(df['alcoholism+depression'][i]==1):
        intervention_arr.append(random.choices(choices_alc_dep_only,weights=weights)[0])
      if(df['alcoholism+tobacco'][i]==1):
        intervention_arr.append(random.choices(choices_alc_tobacco_only,weights=weights)[0])
      if(df['depression+tobacco'][i]==1):
        intervention_arr.append(random.choices(choices_dep_tobacco_only,weights=weights)[0])
      if(df['tobacco+alcohol+smoking'][i]==1):
        intervention_arr.append(random.choices(choices_all_3,weights=weights)[0])
      if(df['alcoholism'][i]==0 and df['depression'][i]==0 and df['tobacco'][i]==0):
        intervention_arr.append('UNAFFECTED')
    df['Intervention'] = intervention_arr
    #print(np.unique(intervention_arr,return_councentage_ts=True))
    df['treatment_outcomes'] = " "
    treatment_outcomes_single_ni = []
    treatment_outcomes_two_ni = []
    treatment_outcomes_three_ni = []
    treatment_outcomes_i = []
    #treatment_outcomes_noth = []

    list_noth = list(np.where(df['Intervention'] == 'UNAFFECTED')[0])
    values_noth = np.random.choice(2,len(list_noth),p=[1-treatment_noth,treatment_noth])
    for i in range(len(list_noth)):
      df.loc[list_noth[i],"treatment_outcomes"] = values_noth[i]
    
    list_single_ni = list(np.where((df['Intervention'] == 'NAlc') | (df['Intervention'] == 'ND') | (df['Intervention'] == 'NT'))[0])
    values_single_ni = np.random.choice(2,len(list_single_ni),p=[1-treatment_1_conditions,treatment_1_conditions])
    for i in range(len(list_single_ni)):
      df.loc[list_single_ni[i],"treatment_outcomes"] = values_single_ni[i]

    list_two_ni = list(np.where((df['Intervention'] == 'NAD') | (df['Intervention'] == 'NDT') | (df['Intervention'] == 'NAT'))[0])
    values_two_ni = np.random.choice(2,len(list_two_ni),p=[1-treatment_2_conditions,treatment_2_conditions])
    for i in range(len(list_two_ni)):
      df.loc[list_two_ni[i],"treatment_outcomes"] = values_two_ni[i]

    list_three_ni = list(np.where(df['Intervention'] == 'NADT')[0])
    values_three_ni = np.random.choice(2,len(list_three_ni),p=[1-treatment_3_conditions,treatment_3_conditions])
    for i in range(len(list_three_ni)):
      df.loc[list_three_ni[i],"treatment_outcomes"] = values_three_ni[i]    
    list_single_inter=list(np.where((df['Intervention'] == 'A') | (df['Intervention'] == 'D') | (df['Intervention'] == 'T'))[0])
    s_int=(1-treatment_1_conditions)/2
    values_single_inter = np.random.choice(2,len(list_single_inter),p=[s_int,treatment_1_conditions+s_int])
    for i in range(len(list_single_inter)):
      df.loc[list_single_inter[i],"treatment_outcomes"] = values_single_inter[i]

    list_double_inter=list(np.where((df['Intervention'] == 'AD') | (df['Intervention'] == 'DT') | (df['Intervention'] == 'AT'))[0])
    d_int=(1-treatment_2_conditions)/2
    values_double_inter = np.random.choice(2,len(list_double_inter),p=[d_int,treatment_2_conditions + d_int])
    for i in range(len(list_double_inter)):
      df.loc[list_double_inter[i],"treatment_outcomes"] = values_double_inter[i]

    list_triple_inter=list(np.where((df['Intervention'] == 'ADT'))[0])
    t_int = (1-treatment_3_conditions)/2
    values_triple_inter = np.random.choice(2,len(list_triple_inter),p=[t_int,treatment_3_conditions+ t_int])
    for i in range(len(list_triple_inter)):
      df.loc[list_triple_inter[i],"treatment_outcomes"] = values_triple_inter[i]


    return StreamingResponse(
        iter([df.to_csv(index=False)]),
        media_type="text/csv",
        headers={"Content-Disposition": f"attachment; filename=init_dataset.csv"})
  except Exception as ex:
    print(ex)

def create_dataset(size,percentage_alc_only, percentage_dep_only, percentage_tobacco_only,percentage_alc_dep, percentage_alc_tobacco,percentage_dep_tobacco, 
                    percentage_tobacco_alcoholism_depression,treatment_noth,treatment_1_conditions,treatment_2_conditions,treatment_3_conditions,
                    gender,bmi,edu,seed,age):
  """
  User Defined Inputs:
  1.size=Population Sample size
  2.percentage_alcoholism/depression/tobacco=Percentage of the the respective condition in the population
  3.percentage_alcoholism_depression(and other dual variable overlaps)= Percentage of the overlap of respective two condtions in the populations
  4.percentage_tobacco_alcoholism_depression= Percentage of the overlap of all 3 cases in the population
  5.gender, bmi, edu: A list consisting of the percentages of distribution in the various categorical buckets for the following variables:
  eg;  
  buckets; gender: 2 buckets; bmi: 3 buckets; edu: 4 buckets
  6.age has 4 parameters - a_min, a_max, mean, std dev
  6.Treatment_noth/treatment_1_conditions=The percentage of good treatment outcomes for the population with the following number of conditions
  """
  zeros=1-(percentage_alc_only + percentage_dep_only  + percentage_tobacco_only + percentage_alc_dep + percentage_alc_tobacco + percentage_dep_tobacco + percentage_tobacco_alcoholism_depression)
  
  print("zeros:{}".format(zeros))
  np.random.seed(2020)
  condition=np.random.choice(8,size,p=[zeros, percentage_alc_only, percentage_dep_only,percentage_alc_dep,percentage_tobacco_only,percentage_alc_tobacco, percentage_dep_tobacco, percentage_tobacco_alcoholism_depression])
  alcoholism=np.where((condition == 1) | (condition==3) | (condition==5) | (condition==7),1,0)
  depression=np.where((condition == 2) | (condition==3) | (condition==6) | (condition==7),1,0)
  
  alc_only=np.where(condition == 1,1,0)
  dep_only=np.where(condition == 2,1,0)
  tobacco_only=np.where(condition == 4,1,0)

  alc_dep=np.where(condition == 3,1,0)
  alc_tobacco=np.where(condition == 5,1,0)
  dep_tobacco=np.where(condition == 6,1,0)

  tobacco=np.where((condition == 4) | (condition==5) | (condition==6) | (condition==7),1,0)
  all_three=np.where(condition == 7,1,0)
  age_1= np.random.normal(loc=age[2], scale=age[3], size=size)
  age_1=np.clip(age_1, a_min=age[0], a_max=age[1])
  age_1 = np.round(age_1).astype(int)
  sex=np.random.choice(2,size,p=[gender[0],gender[1]])
  body_mass=np.random.choice(3,size,p=[bmi[0],bmi[1],bmi[2]])
  education=np.random.choice(4,size,p=[edu[0],edu[1],edu[2],edu[3]])
  cavitation = np.random.choice(2,size,p=[0.5,0.5])
  ttd = np.random.normal(loc=7, scale=3, size=size)
  df = pd.DataFrame(
    {
        'idx': np.arange(1, size+1),
        'age': age_1,
        'gender': sex,
        'bmi': body_mass,
        'education': education,
        'cavitation': cavitation,
      	'TTD': ttd,
        'alcoholism': alcoholism,
        'depression': depression,
        'tobacco': tobacco,
        'alcohol_only':alc_only,
        'depression_only':dep_only,
        'tobacco_only': tobacco_only,
        'alcoholism+depression':alc_dep,
        'alcoholism+tobacco':alc_tobacco,
        'depression+tobacco':dep_tobacco,
        'tobacco+alcohol+smoking':all_three
     }
  )
  intervention_arr=[]
  choices_alc_only=['NAlc','A']
  choices_dep_only=['ND','D']
  choices_tobacco_only=['NT','T']
  choices_alc_dep_only=['NAD','AD']
  choices_alc_tobacco_only=['NAT','AT']
  choices_dep_tobacco_only=['NDT','DT']
  choices_all_3=['NADT','ADT']
  random.seed(seed)
  weights=[0.5,0.5]
  for i in range(size):
    if(df['alcohol_only'][i]==1):
      #intervention_arr.append(np.random.binomial(1,0.5,size=1))
      intervention_arr.append(random.choices(choices_alc_only,weights=weights)[0])
    if(df['depression_only'][i]==1):
      intervention_arr.append(random.choices(choices_dep_only,weights=weights)[0])
    if(df['tobacco_only'][i]==1):
      intervention_arr.append(random.choices(choices_tobacco_only,weights=weights)[0])
    if(df['alcoholism+depression'][i]==1):
      intervention_arr.append(random.choices(choices_alc_dep_only,weights=weights)[0])
    if(df['alcoholism+tobacco'][i]==1):
      intervention_arr.append(random.choices(choices_alc_tobacco_only,weights=weights)[0])
    if(df['depression+tobacco'][i]==1):
      intervention_arr.append(random.choices(choices_dep_tobacco_only,weights=weights)[0])
    if(df['tobacco+alcohol+smoking'][i]==1):
      intervention_arr.append(random.choices(choices_all_3,weights=weights)[0])
    if(df['alcoholism'][i]==0 and df['depression'][i]==0 and df['tobacco'][i]==0):
      intervention_arr.append('UNAFFECTED')
  df['Intervention'] = intervention_arr
  #print(np.unique(intervention_arr,return_councentage_ts=True))
  df['treatment_outcomes'] = " "
  treatment_outcomes_single_ni = []
  treatment_outcomes_two_ni = []
  treatment_outcomes_three_ni = []
  treatment_outcomes_i = []
  #treatment_outcomes_noth = []

  list_noth = list(np.where(df['Intervention'] == 'UNAFFECTED')[0])
  values_noth = np.random.choice(2,len(list_noth),p=[1-treatment_noth,treatment_noth])
  for i in range(len(list_noth)):
    df.loc[list_noth[i],"treatment_outcomes"] = values_noth[i]
  
  list_single_ni = list(np.where((df['Intervention'] == 'NAlc') | (df['Intervention'] == 'ND') | (df['Intervention'] == 'NT'))[0])
  values_single_ni = np.random.choice(2,len(list_single_ni),p=[1-treatment_1_conditions,treatment_1_conditions])
  for i in range(len(list_single_ni)):
    df.loc[list_single_ni[i],"treatment_outcomes"] = values_single_ni[i]

  list_two_ni = list(np.where((df['Intervention'] == 'NAD') | (df['Intervention'] == 'NDT') | (df['Intervention'] == 'NAT'))[0])
  values_two_ni = np.random.choice(2,len(list_two_ni),p=[1-treatment_2_conditions,treatment_2_conditions])
  for i in range(len(list_two_ni)):
    df.loc[list_two_ni[i],"treatment_outcomes"] = values_two_ni[i]

  list_three_ni = list(np.where(df['Intervention'] == 'NADT')[0])
  values_three_ni = np.random.choice(2,len(list_three_ni),p=[1-treatment_3_conditions,treatment_3_conditions])
  for i in range(len(list_three_ni)):
    df.loc[list_three_ni[i],"treatment_outcomes"] = values_three_ni[i]

  #list_i = list(np.where((df['Intervention'] == 'A') | (df['Intervention'] == 'D') | (df['Intervention'] == 'T') | (df['Intervention'] == 'AD') | (df['Intervention'] == 'AT') | (df['Intervention'] == 'DT') | (df['Intervention'] == 'ADT'))[0])
  #values_i = np.random.choice(2,len(list_i),p=[1-treatment_intervention,treatment_intervention])
  #for i in range(len(list_i)):
   # df.loc[list_i[i],"treatment_outcomes"] = values_i[i]
  
  list_single_inter=list(np.where((df['Intervention'] == 'A') | (df['Intervention'] == 'D') | (df['Intervention'] == 'T'))[0])
  s_int=(1-treatment_1_conditions)/2
  values_single_inter = np.random.choice(2,len(list_single_inter),p=[s_int,treatment_1_conditions+s_int])
  for i in range(len(list_single_inter)):
    df.loc[list_single_inter[i],"treatment_outcomes"] = values_single_inter[i]

  list_double_inter=list(np.where((df['Intervention'] == 'AD') | (df['Intervention'] == 'DT') | (df['Intervention'] == 'AT'))[0])
  d_int=(1-treatment_2_conditions)/2
  values_double_inter = np.random.choice(2,len(list_double_inter),p=[d_int,treatment_2_conditions + d_int])
  for i in range(len(list_double_inter)):
    df.loc[list_double_inter[i],"treatment_outcomes"] = values_double_inter[i]

  list_triple_inter=list(np.where((df['Intervention'] == 'ADT'))[0])
  t_int = (1-treatment_3_conditions)/2
  values_triple_inter = np.random.choice(2,len(list_triple_inter),p=[t_int,treatment_3_conditions+ t_int])
  for i in range(len(list_triple_inter)):
    df.loc[list_triple_inter[i],"treatment_outcomes"] = values_triple_inter[i]

  return df


@app.post('/calc_power')
def calculate_statistical_power(x:str,y:str,file:UploadFile=File(...)):
    """ x = value of treatment variable (gives us idea of the population - NAlc, a, nt, t, adt, etc)
        y = value of control variable """
    try:
        df = pd.read_csv(file.file)
        file.file.close
        #print(df.head())
        power_analysis=TTestIndPower()
        treatment_arr=[]
        control_arr=[]
        treatment_locs=np.where((df['Intervention']==x))
        control_locs=np.where((df['Intervention']==y))
        for i in treatment_locs:
            treatment_arr.append(df['treatment_outcomes'].iloc[i])
        for j in control_locs:
            control_arr.append(df['treatment_outcomes'].iloc[j])
        l1 = len(treatment_arr[0])
        l2 = len(control_arr[0])
        index_treatment=np.arange(0,l1)
        index_control=np.arange(0,l2)
        treatment_df=pd.DataFrame({'idx':index_treatment,"Treatment":treatment_arr[0]})
        control_df=pd.DataFrame({'idx':index_control,"Control":control_arr[0]}) 
        mu1=treatment_df['Treatment'].mean()
        mu2=control_df['Control'].mean()
        std1=treatment_df['Treatment'].std()
        std2=control_df['Control'].std()
        s = np.sqrt(((l1 - 1) * std1 + (l2 - 1) * std2) / (l1 + l2 - 2))
        d = (mu1 - mu2) / s #cohen's effect size
        eff = round(d,2)
        p = power_analysis.power(effect_size=eff,alpha=0.05,nobs1=l1,ratio=(l1/l2),alternative='two-sided')
        return p
    except Exception as ex:
      print(ex) 

def calculate_power1(x, y, df):
    power_analysis=TTestIndPower()
    treatment_arr=[]
    control_arr=[]
    treatment_locs=np.where((df['Intervention']==x))
    control_locs=np.where((df['Intervention']==y))
    for i in treatment_locs:
        treatment_arr.append(df['treatment_outcomes'].iloc[i])
    for j in control_locs:
        control_arr.append(df['treatment_outcomes'].iloc[j])
    l1 = len(treatment_arr[0])
    l2 = len(control_arr[0])
    index_treatment=np.arange(0,l1)
    index_control=np.arange(0,l2)
    treatment_df=pd.DataFrame({'idx':index_treatment,"Treatment":treatment_arr[0]})
    control_df=pd.DataFrame({'idx':index_control,"Control":control_arr[0]}) 
    mu1=treatment_df['Treatment'].mean()
    mu2=control_df['Control'].mean()
    std1=treatment_df['Treatment'].std()
    std2=control_df['Control'].std()
    s = np.sqrt(((l1 - 1) * std1 + (l2 - 1) * std2) / (l1 + l2 - 2))
    d = (mu1 - mu2) / s #cohen's effect size
    eff = round(d,2)
    p = power_analysis.power(effect_size=eff,alpha=0.05,nobs1=l1,ratio=(l1/l2),alternative='two-sided')
    return p

#@app.post('/check_samp_sizes')
def check_sample_sizes(dfx):
  sample_sizes={}
  sample_sizes['Alcohol']=calculate_power1('A','NAlc',dfx)
  sample_sizes['Depression']=calculate_power1('D','ND',dfx)  
  sample_sizes['Tobacco']=calculate_power1('T','NT',dfx)  
  sample_sizes['Alcohol-Depression']=calculate_power1('AD','NAD',dfx)  
  sample_sizes['Alcohol-Tobacco']=calculate_power1('AT','NAT',dfx)  
  sample_sizes['Depression-Tobacco']=calculate_power1('DT','NDT',dfx) 
  sample_sizes['Alcohol-Depression-Tobacco']=calculate_power1('ADT','NADT',dfx) 
  #for i in sample_sizes.values():
  all_above_threshold = all(value >= 0.8 for value in sample_sizes.values())
  print(all_above_threshold)
  print(sample_sizes.values())
  return sample_sizes.values
# check_sample_sizes(df_2)

def try_sample_sizes3(dfx):
    sample_sizes = {}
    sample_sizes['Alcohol'] = calculate_power1('A', 'NAlc', dfx)
    sample_sizes['Depression'] = calculate_power1('D', 'ND', dfx)
    sample_sizes['Tobacco'] = calculate_power1('T', 'NT', dfx)
    sample_sizes['Alcohol-Depression'] = calculate_power1('AD', 'NAD', dfx)
    sample_sizes['Alcohol-Tobacco'] = calculate_power1('AT', 'NAT', dfx)
    sample_sizes['Depression-Tobacco'] = calculate_power1('DT', 'NDT', dfx)
    sample_sizes['Alcohol-Depression-Tobacco'] = calculate_power1('ADT', 'NADT', dfx)

    size = dfx.shape[0]
    #total_power = sum(sample_sizes.values())
    all_above_threshold = all(value >= 0.79 for value in sample_sizes.values())
    if not all_above_threshold:
        if sample_sizes['Alcohol'] < 0.79:
            size += 100
        elif sample_sizes['Depression'] < 0.79:
            size += 100
        elif sample_sizes['Tobacco'] < 0.79:
            size += 100
        elif sample_sizes['Alcohol-Depression'] < 0.79:
            size += 80
        elif sample_sizes['Alcohol-Tobacco'] < 0.79:
            size += 80
        elif sample_sizes['Depression-Tobacco'] < 0.79:
            size += 80
        elif sample_sizes['Alcohol-Depression-Tobacco'] < 0.79:
            size += 20
          
        ratio_alc_only = round(dfx['Intervention'].value_counts(normalize=True)['A']+dfx['Intervention'].value_counts(normalize=True)['NAlc'],2)
        ratio_dep_only = round(dfx['Intervention'].value_counts(normalize=True)['D']+dfx['Intervention'].value_counts(normalize=True)['ND'],2)
        ratio_tob_only = round(dfx['Intervention'].value_counts(normalize=True)['T']+dfx['Intervention'].value_counts(normalize=True)['NT'],2)
        ratio_at = round(dfx['Intervention'].value_counts(normalize=True)['AT']+dfx['Intervention'].value_counts(normalize=True)['NAT'],2)
        ratio_ad = round(dfx['Intervention'].value_counts(normalize=True)['AD']+dfx['Intervention'].value_counts(normalize=True)['NAD'],2)
        ratio_dt = round(dfx['Intervention'].value_counts(normalize=True)['DT']+dfx['Intervention'].value_counts(normalize=True)['NDT'],2)
        ratio_adt = round(dfx['Intervention'].value_counts(normalize=True)['ADT']+dfx['Intervention'].value_counts(normalize=True)['NADT'],2)
        
        #dfx = create_dataset2(size, 0.08, 0.08, 0.08, 0.04, 0.04, 0.04, 0.03, 0.9, 0.80, 0.70, 0.60, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],seed=52)
        dfx = create_dataset(size, ratio_alc_only, ratio_dep_only, ratio_tob_only, ratio_ad, ratio_at, ratio_dt, ratio_adt, 0.95, 0.90, 0.85, 0.80, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],age=[18,60,35,15],seed=52)
        return try_sample_sizes3(dfx)
    while all(value > 0.85 for value in sample_sizes.values()):
            print('inside shrinking')
                #for group in sample_sizes.keys:
                    #if sample_sizes[group] > 0.95:
                    #   size -= 1500
            if sample_sizes['Alcohol'] > 0.95:
                size -= 1500 
            elif sample_sizes['Tobacco'] > 0.95:
                size -= 1500
            elif sample_sizes['Depression'] > 0.95:
                size -= 1500
            ratio_alc_only = round(dfx['Intervention'].value_counts(normalize=True)['A']+dfx['Intervention'].value_counts(normalize=True)['NAlc'],2)
            ratio_dep_only = round(dfx['Intervention'].value_counts(normalize=True)['D']+dfx['Intervention'].value_counts(normalize=True)['ND'],2)
            ratio_tob_only = round(dfx['Intervention'].value_counts(normalize=True)['T']+dfx['Intervention'].value_counts(normalize=True)['NT'],2)
            ratio_at = round(dfx['Intervention'].value_counts(normalize=True)['AT']+dfx['Intervention'].value_counts(normalize=True)['NAT'],2)
            ratio_ad = round(dfx['Intervention'].value_counts(normalize=True)['AD']+dfx['Intervention'].value_counts(normalize=True)['NAD'],2)
            ratio_dt = round(dfx['Intervention'].value_counts(normalize=True)['DT']+dfx['Intervention'].value_counts(normalize=True)['NDT'],2)
            ratio_adt = round(dfx['Intervention'].value_counts(normalize=True)['ADT']+dfx['Intervention'].value_counts(normalize=True)['NADT'],2)
                
            dfx = create_dataset(size, ratio_alc_only, ratio_dep_only, ratio_tob_only, ratio_ad, ratio_at, ratio_dt, ratio_adt, 0.95, 0.90, 0.85, 0.80, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],age=[18,60,35,15],seed=52)
            #dfx = create_dataset2(size, 0.08, 0.08, 0.08, 0.04, 0.04, 0.04, 0.03, 0.9, 0.80, 0.70, 0.60, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],seed=52)
            return try_sample_sizes3(dfx)
    return dfx

@app.post('/try_samp_sizes')
def Optimize_sample_size(file:UploadFile=File(...)):
    """
    Accepts: Inital dataset csv file
    Returns: Dataset with optimal power for the given sample size(Optimized sample size)
    Employs a recursive resampling algorithm
    """
    try:
        dfx = pd.read_csv(file.file)
        file.file.close
        #dfx = pd.read_csv(io.BytesIO(contents), encoding='utf-8')
        #file.file.close
        sample_sizes = {}
        sample_sizes['Alcohol'] = calculate_power1('A', 'NAlc', dfx)
        sample_sizes['Depression'] = calculate_power1('D', 'ND', dfx)
        sample_sizes['Tobacco'] = calculate_power1('T', 'NT', dfx)
        sample_sizes['Alcohol-Depression'] = calculate_power1('AD', 'NAD', dfx)
        sample_sizes['Alcohol-Tobacco'] = calculate_power1('AT', 'NAT', dfx)
        sample_sizes['Depression-Tobacco'] = calculate_power1('DT', 'NDT', dfx)
        sample_sizes['Alcohol-Depression-Tobacco'] = calculate_power1('ADT', 'NADT', dfx)

        size = dfx.shape[0]
        #total_power = sum(sample_sizes.values())
        all_above_threshold = all(value >= 0.79 for value in sample_sizes.values())
        if not all_above_threshold:
            if sample_sizes['Alcohol'] < 0.79:
                size += 100
            elif sample_sizes['Depression'] < 0.79:
                size += 100
            elif sample_sizes['Tobacco'] < 0.79:
                size += 100
            elif sample_sizes['Alcohol-Depression'] < 0.79:
                size += 80
            elif sample_sizes['Alcohol-Tobacco'] < 0.79:
                size += 80
            elif sample_sizes['Depression-Tobacco'] < 0.79:
                size += 80
            elif sample_sizes['Alcohol-Depression-Tobacco'] < 0.79:
                size += 20
              
            ratio_alc_only = round(dfx['Intervention'].value_counts(normalize=True)['A']+dfx['Intervention'].value_counts(normalize=True)['NAlc'],2)
            ratio_dep_only = round(dfx['Intervention'].value_counts(normalize=True)['D']+dfx['Intervention'].value_counts(normalize=True)['ND'],2)
            ratio_tob_only = round(dfx['Intervention'].value_counts(normalize=True)['T']+dfx['Intervention'].value_counts(normalize=True)['NT'],2)
            ratio_at = round(dfx['Intervention'].value_counts(normalize=True)['AT']+dfx['Intervention'].value_counts(normalize=True)['NAT'],2)
            ratio_ad = round(dfx['Intervention'].value_counts(normalize=True)['AD']+dfx['Intervention'].value_counts(normalize=True)['NAD'],2)
            ratio_dt = round(dfx['Intervention'].value_counts(normalize=True)['DT']+dfx['Intervention'].value_counts(normalize=True)['NDT'],2)
            ratio_adt = round(dfx['Intervention'].value_counts(normalize=True)['ADT']+dfx['Intervention'].value_counts(normalize=True)['NADT'],2)
            
            #dfx = create_dataset2(size, 0.08, 0.08, 0.08, 0.04, 0.04, 0.04, 0.03, 0.9, 0.80, 0.70, 0.60, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],seed=52)
            dfx = create_dataset(size, ratio_alc_only, ratio_dep_only, ratio_tob_only, ratio_ad, ratio_at, ratio_dt, ratio_adt, 0.95, 0.90, 0.85, 0.80, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],seed=52,age=[18,60,35,15])
            return try_sample_sizes3(dfx)
        while all(value > 0.85 for value in sample_sizes.values()):
                print('inside shrinking')
                #for group in sample_sizes.keys:
                    #if sample_sizes[group] > 0.95:
                    #   size -= 1500
                if sample_sizes['Alcohol'] > 0.95:
                  size -= 1500 
                elif sample_sizes['Tobacco'] > 0.95:
                  size -= 1500
                elif sample_sizes['Depression'] > 0.95:
                  size -= 1500
                    #if any(value < 0.80 for value in sample_sizes.values()):
                
                ratio_alc_only = round(dfx['Intervention'].value_counts(normalize=True)['A']+dfx['Intervention'].value_counts(normalize=True)['NAlc'],2)
                ratio_dep_only = round(dfx['Intervention'].value_counts(normalize=True)['D']+dfx['Intervention'].value_counts(normalize=True)['ND'],2)
                ratio_tob_only = round(dfx['Intervention'].value_counts(normalize=True)['T']+dfx['Intervention'].value_counts(normalize=True)['NT'],2)
                ratio_at = round(dfx['Intervention'].value_counts(normalize=True)['AT']+dfx['Intervention'].value_counts(normalize=True)['NAT'],2)
                ratio_ad = round(dfx['Intervention'].value_counts(normalize=True)['AD']+dfx['Intervention'].value_counts(normalize=True)['NAD'],2)
                ratio_dt = round(dfx['Intervention'].value_counts(normalize=True)['DT']+dfx['Intervention'].value_counts(normalize=True)['NDT'],2)
                ratio_adt = round(dfx['Intervention'].value_counts(normalize=True)['ADT']+dfx['Intervention'].value_counts(normalize=True)['NADT'],2)
                    
                dfx = create_dataset(size, ratio_alc_only, ratio_dep_only, ratio_tob_only, ratio_ad, ratio_at, ratio_dt, ratio_adt, 0.95, 0.90, 0.85, 0.80, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],seed=52,age=[18,60,35,15])
                #dfx = create_dataset2(size, 0.08, 0.08, 0.08, 0.04, 0.04, 0.04, 0.03, 0.9, 0.80, 0.70, 0.60, gender=[0.5, 0.5], bmi=[0.2,0.5,0.3],edu=[0.1,0.2,0.2,0.5],seed=52)
                return try_sample_sizes3(dfx)
        #return dfx
        return StreamingResponse(
                    iter([dfx.to_csv(index=False)]),
                    media_type="text/csv",
                    headers={"Content-Disposition": f"attachment; filename=resim_screening.csv"})
    except Exception as ex:
      print(ex) 

@app.post('/clinical_dataset')
def clinical_dataset(file:UploadFile=File(...)):
    """
    Function accepts an optimized sample size dataset and returns a clinical dataset consisting of equal number of samples from each population group.
    """
    try:
        dfx = pd.read_csv(file.file)
        file.file.close
        counts= dfx['Intervention'].value_counts()[1]

        max_unaffected= dfx['Intervention'].value_counts()[0]
        column_name = 'Intervention'

        condition = dfx[column_name] == 'UNAFFECTED'
        selected_rows = dfx[condition]

        new_value = 'unaffected'
        count_to_reduce = max_unaffected - (max_unaffected-counts)
        count_reduced = 0

        for index, row in selected_rows.iterrows():
            if count_reduced < count_to_reduce:
                dfx.at[index, column_name] = new_value
                count_reduced += 1
            else:
                break
        condition = dfx['Intervention'] != 'UNAFFECTED'
        clinical = dfx[condition]

        clinical = clinical.dropna()
        #return clinical
        return StreamingResponse(
                    iter([clinical.to_csv(index=False)]),
                    media_type="text/csv",
                    headers={"Content-Disposition": f"attachment; filename=clinical_dataset.csv"})
    except Exception as ex:
        print(ex)

if __name__ == '__main__':
    uvicorn.run(app, host='127.0.0.1', port=8000)