Initial commit

app.py

@@ -0,0 +1,90 @@
+import gradio as gr
+import numpy as np
+from PIL import Image
+import requests
+import pandas as pd
+import hopsworks
+import joblib
+
+#connect to hopsworks
+project = hopsworks.login(project="test42")
+fs = project.get_feature_store()
+
+#get model
+mr = project.get_model_registry()#connect to model registry
+model = mr.get_model("titanic_model_modal", version=1) #retrieve model from hopsworks
+model_dir = model.download() #download model to cur dir
+model = joblib.load(model_dir + "/titanic_model.pkl") #load model from cur dir
+
+
+def passenger(pclass,#index
+              age,#float
+              sibsp,#float
+              parch,#float
+              fare,#float
+              sex,#index 0-male,1-female
+              deck,# index abcdefgnt
+              embarked# index cnqs
+              ):
+    deck_all="abcdefgnt"
+    embarked_all="cnqs"
+    deck_count=[0 for i in deck_all]
+    deck_count[deck]=1
+    embarked_count=[0 for i in embarked_all]
+    embarked_count[embarked]=1
+    
+    input_df = pd.DataFrame({"pclass":[pclass+1],
+              "age":[age],
+              "sibsp":[sibsp],
+              "parch":[parch],
+              "fare":[fare],
+              "sex_female":[sex],
+              "sex_male":[1-sex],
+              "deck_a":deck_count[0],
+              "deck_b":deck_count[1],
+              "deck_c":deck_count[2],
+              "deck_d":deck_count[3],
+              "deck_e":deck_count[4],
+              "deck_f":deck_count[5],
+              "deck_g":deck_count[6],
+              "deck_n":deck_count[7],
+              "deck_t":deck_count[8],
+              "embarked_c":embarked_count[0],
+              "embarked_n":embarked_count[1],
+              "embarked_q":embarked_count[2],
+              "embarked_s":embarked_count[3],
+                  })
+    # 'res' is a list of predictions returned as the label.
+    #print(input_df)
+    res = model.predict(input_df) #prediction from model based on input
+    # We add '[0]' to the result of the transformed 'res', because 'res' is a list, and we only want 
+    # the first element.
+    if res==0:
+        url="https://i.imgflip.com/5jvc2d.jpg"
+    else:
+        url="https://encrypted-tbn3.gstatic.com/images?q=tbn:ANd9GcTv3XuQKvjiF_ZpUt8rKlsVBX--JXMpfa674H03N-aApj_HjK1S"
+    img = Image.open(requests.get(url, stream=True).raw) #get image from github    
+    return img
+
+#create hugging face interface
+demo = gr.Interface(
+    passenger,
+    [
+        gr.Dropdown(["first", "second", "third"], type="index",label="Passenger Class"),
+        gr.Slider(0, 80, value=25,label="Age"),
+        gr.Slider(0, 10, step=1, value=0, label="Number of siblings/spouses aboard the Titanic"),
+        gr.Slider(0, 10, step=1, value=0, label="Number of parents/children aboard the Titanic"),
+        gr.Number(default=0, label="Passenger fare"),
+        gr.Radio(["Male","Female"],type="index",label="Sex"),
+        gr.Radio([f"Deck_{c}" for c in "ABCDEFGNT"],type="index",label="Deck (Select N if unknown)"),
+        gr.Radio([f"Embarked_{e}" for e in "CNQS"],type="index",label="Embark point (Select N if unknown)")
+        
+    ],
+    title="Titanic Survivor Predictive Analytics",
+    description="Who could surive the titanic",
+    allow_flagging="never",
+    outputs=gr.Image(type="pil")
+    )
+
+demo.launch()
+

requirements.txt

@@ -0,0 +1,8 @@
+hopsworks
+pandas
+joblib
+scikit-learn
+seaborn
+dataframe-image
+modal-client
+gradio