File size: 7,233 Bytes
5e5dfdc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2766617
 
 
 
 
5e5dfdc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
import os
import logging
from dotenv import load_dotenv
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
import openai
import numpy as np
import pandas as pd
from sklearn.linear_model import LinearRegression
import requests

# Load environment variables from the .env file
load_dotenv()

# Set the OpenAI API key
openai.api_key = os.getenv("OPENAI_API_KEY")

# File paths
predefined_constructs_file = "/app/predefined_constructs.txt"
full_matrix_path = "/app/no_na_matrix.csv"

# Logging configuration
logging.basicConfig(
    level=logging.DEBUG,
    format="%(asctime)s - %(name)s - %(levelname)s - %(message)s",
    handlers=[
        logging.FileHandler("app.log"),
        logging.StreamHandler()
    ]
)

logger = logging.getLogger(__name__)

# FastAPI application
app = FastAPI()

# Root endpoint for health check
@app.get("/")
def read_root():
    return {"message": "API is up and running!"}
    
# Request schema
class InferenceRequest(BaseModel):
    query: str

# Functions for inference
def read_predefined_constructs(file_path):
    with open(file_path, 'r') as file:
        return file.read()

def call_openai_chat(model, messages):
    url = "https://api.openai.com/v1/chat/completions"
    headers = {
        "Content-Type": "application/json",
        "Authorization": f"Bearer {openai.api_key}",
    }
    data = {"model": model, "messages": messages}
    response = requests.post(url, headers=headers, json=data)
    response.raise_for_status()
    return response.json()["choices"][0]["message"]["content"]

def clean_predefined_constructs_and_matrix(predefined_constructs_path, full_matrix_path):
    with open(predefined_constructs_path, "r") as f:
        predefined_constructs = [
            line.split('\t')[0].strip().lower()
            for line in f.readlines() if line.strip()
        ]

    predefined_constructs = [
        c.replace('-', ' ').replace('–', ' ').replace('(', '')
         .replace(')', '').replace('/', ' ').replace(',', ' ')
         .replace('.', ' ').strip()
        for c in predefined_constructs if c != "variable"
    ]

    full_matrix = pd.read_csv(full_matrix_path, index_col=0)
    full_matrix.index = full_matrix.index.str.strip().str.lower().str.replace('[^a-z0-9 ]', '', regex=True)
    full_matrix.columns = full_matrix.columns.str.strip().str.lower().str.replace('[^a-z0-9 ]', '', regex=True)

    valid_constructs = set(predefined_constructs)
    matrix_rows = set(full_matrix.index)
    matrix_columns = set(full_matrix.columns)

    invalid_rows = matrix_rows - valid_constructs
    invalid_columns = matrix_columns - valid_constructs

    full_matrix = full_matrix.drop(index=invalid_rows, errors='ignore')
    full_matrix = full_matrix.drop(columns=invalid_columns, errors='ignore')

    missing_constructs = [c for c in predefined_constructs if c not in full_matrix.index or c not in full_matrix.columns]
    if missing_constructs:
        raise ValueError(f"Missing constructs in the correlation matrix: {missing_constructs}")

    return predefined_constructs, full_matrix

def construct_analyzer(user_prompt, predefined_constructs):
    predefined_constructs_prompt = "\n".join(predefined_constructs)
    prompt_text = (
        f"Here is the user's prompt: '{user_prompt}'.\n\n"
        "Your role is to identify 10 relevant constructs from the provided list. "
        "Pick one variable as the dependent variable. Format output as:\n"
        "var1|var2|...|var10;dependent_var\n\n"
        f"{predefined_constructs_prompt}"
    )

    messages = [
        {"role": "system", "content": "You are a construct analyzer."},
        {"role": "user", "content": prompt_text},
    ]
    constructs_with_dependent = call_openai_chat(model="gpt-4", messages=messages).strip()
    constructs_list, dependent_variable = constructs_with_dependent.split(';')
    constructs = constructs_list.split('|')

    if dependent_variable not in constructs:
        constructs.append(dependent_variable)

    return constructs, dependent_variable

def regression_analyst(correlation_matrix, dependent_variable):
    normalized_dependent = dependent_variable.strip().lower()
    normalized_columns = [col.strip().lower() for col in correlation_matrix.columns]

    if normalized_dependent not in normalized_columns:
        from difflib import get_close_matches
        suggestions = get_close_matches(normalized_dependent, normalized_columns, n=3, cutoff=0.6)
        raise KeyError(
            f"Dependent variable '{dependent_variable}' not found. Suggestions: {suggestions}"
        )

    original_dependent = correlation_matrix.columns[normalized_columns.index(normalized_dependent)]
    independent_vars = [var for var in correlation_matrix.columns if var != original_dependent]

    synthetic_data = np.random.multivariate_normal(
        mean=np.zeros(len(correlation_matrix)),
        cov=correlation_matrix.to_numpy(),
        size=1000
    )
    synthetic_data = pd.DataFrame(synthetic_data, columns=correlation_matrix.columns)
    X = synthetic_data[independent_vars]
    y = synthetic_data[original_dependent]

    model = LinearRegression()
    model.fit(X, y)
    beta_weights = model.coef_

    return independent_vars, beta_weights

def generate_inference(user_query, equation, independent_vars, dependent_variable, beta_weights):
    beta_details = "\n".join([f"{var}: {round(beta, 4)}" for var, beta in zip(independent_vars, beta_weights)])
    prompt = (
        f"User query: '{user_query}'\n"
        f"Regression Equation: {equation}\n"
        f"Variables and Beta Weights:\n{beta_details}\n\n"
        "Provide actionable insights based on this analysis."
    )

    messages = [
        {"role": "system", "content": "You are a skilled analyst interpreting regression results."},
        {"role": "user", "content": prompt},
    ]
    return call_openai_chat(model="gpt-4", messages=messages).strip()

def run_inference_pipeline(user_query, predefined_constructs, correlation_matrix):
    constructs_raw, dependent_variable = construct_analyzer(user_query, predefined_constructs)
    constructs_list = constructs_raw.split('|')

    if dependent_variable not in constructs_list:
        constructs_list.append(dependent_variable)

    correlation_matrix_filtered = correlation_matrix.loc[constructs_list, constructs_list]
    independent_vars, beta_weights = regression_analyst(correlation_matrix_filtered, dependent_variable)

    equation = f"{dependent_variable} = " + " + ".join(
        [f"{round(beta, 4)}*{var}" for beta, var in zip(beta_weights, independent_vars)]
    )

    inference = generate_inference(user_query, equation, independent_vars, dependent_variable, beta_weights)

    return {"equation": equation, "inference": inference}

# API endpoint
@app.post("/infer")
def infer(request: InferenceRequest):
    try:
        predefined_constructs, cleaned_matrix = clean_predefined_constructs_and_matrix(predefined_constructs_file, full_matrix_path)
        results = run_inference_pipeline(request.query, predefined_constructs, cleaned_matrix)
        return results
    except Exception as e:
        logger.error(f"Error during inference: {e}")
        raise HTTPException(status_code=500, detail=str(e))