import numpy as np
import pickle 
import pandas as pd
from sklearn.preprocessing import LabelEncoder

data = pd.read_csv("funding/follow_on_final.csv")
train = data
train = train.drop(["Organization Name","Equity Only Funding", "Announced Date"], axis = 1 )

le=LabelEncoder()
cols = train.select_dtypes(include=['object']).columns.tolist()

cat_labels = {}
for col in cols:
    train[col] = le.fit_transform(train[col]) 
    cat_labels[col] = list(le.classes_)


def str_to_label(col, val):
    if col in ["Money Raised" , "Total Funding", "Number of Founders"]:
        return float(val)
    return cat_labels[col].index(val) + 1    

model =  pickle.load(open('funding/classifier_model.pkl', 'rb'))


def make_predictions(inputs):
    out = {}

    in_data = []
    for key in inputs:
        in_data.append(str_to_label(key, inputs[key]))

    input_data = np.array(in_data).reshape(-1, len(in_data))

    pred = model.predict_proba(input_data).max()
    if pred > 0.5:
        out["Prediction"] = "Funded"
        out["confidence"] = round(pred * 100)
    else:
         out["Prediction"] = "Not Funded"
         out["confidence"] = round(pred * 100)
        
    return out