initial commit

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+assets/insurance_regression_model.keras filter=lfs diff=lfs merge=lfs -text
+assets/column_transformer.joblib filter=lfs diff=lfs merge=lfs -text

README.md

@@ -1,6 +1,6 @@
 ---
 title: Insurance Regression
-emoji: 🌍
+emoji: 🏥
 colorFrom: green
 colorTo: blue
 sdk: gradio

app.py

@@ -0,0 +1,67 @@
+import numpy as np
+import pandas as pd
+from joblib import load
+import gradio as gr
+from tensorflow.keras.models import load_model
+
+
+# filepaths
+column_transform_path = "assets/column_transformer.joblib" # change during the creation of the huggingface app
+model_path = "assets/insurance_regression_model.keras" # change during the creation of the huggingface app
+
+# instance the models
+column_transformer = load(column_transform_path)
+insurance_model = load_model(model_path)
+
+# short description of the stuff
+title = "Insurance Regressor Model"
+description = "POC of a deep learning regression model for estimating insurance charges based on 11 input variables."
+article = "The model was implemented in tensorflow. It is simple as i made it as a refresher for reviewing the Tensorflow fundamentals (the problem most likely could be solved using classical regression). The dataset used was [Medical Cost Personal Datasets](https://www.kaggle.com/datasets/mirichoi0218/insurance)."
+
+# processing function
+def insurance_prediction(age,gender,bmi,children,smoker,region):
+  columns = ['age', 'sex', 'bmi', 'children', 'smoker', 'region']
+  # transform some data to the datatypes compatible with column transfomer
+  age = int(age)
+  gender = "male" if gender else "female"
+  children = int(children)
+  smoker = "yes" if smoker else "no"
+  region = region[0] if len(region) > 0 else 'northeast' # since it is a proof of concept, i left the value there.
+  # arange the values into a list
+  values = [age,gender,bmi,children,smoker,region]
+  # create a dictionary to structure the sample at hand
+  sample = {k:v for k,v in zip(columns,values)}
+  # create a dataframe using the sample 
+  X = pd.DataFrame(data = pd.Series(sample)).T
+  # preprocess the dataframe
+  X_preprocessed = column_transformer.transform(X)
+  # make the prediction accordingly 
+  y_pred = insurance_model.predict(X_preprocessed)
+  return y_pred[0][0]
+
+
+
+
+
+demo = gr.Interface(
+    fn = insurance_prediction,
+    inputs = [
+        gr.Number(value = 40,label = "Age",show_label=True),
+        gr.Checkbox(value = False,label = "Male",show_label = True),
+        gr.Number(value = 30,label = "Body Mass Index",show_label=True),
+        gr.Number(value = 1,label = "Number of Children",show_label=True),
+        gr.Checkbox(value = False,label = "Smoker",show_label = True),
+        gr.CheckboxGroup(choices = ['northeast', 'northwest', 'southeast', 'southwest'],
+                         value = ['northeast', 'northwest', 'southeast', 'southwest'],
+                         label="Region",
+                         info="Will only take into consideration the fist one selected.")
+        ],
+    outputs = [
+        gr.Number(label = "Insurance Charges Estimate (American Dollars)")
+        ],
+    title = title,
+    description = description,
+    article = article
+)
+
+demo.launch()

assets/column_transformer.joblib

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d53d73e17da5e8fb4822f307772383138631fa489d251d7844071d87e5a0a0e2
+size 3503

assets/insurance_regression_model.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:291424b48786654966fe4dd77332444458cd5e33793872ea1bec708b9db3ccd8
+size 78752

requirements.txt

@@ -0,0 +1,6 @@
+tensorflow==2.12.0
+gradio==3.39.0
+pandas==1.5.3
+numpy==1.23.5
+scikit-learn==1.2.2
+joblib==1.3.1