app.py

import os
os.system('pip install openpyxl')
os.system('pip install sentence-transformers')
import pandas as pd
import gradio as gr
import statistics
from sklearn.neighbors import NearestNeighbors
from sentence_transformers import SentenceTransformer

df = pd.read_parquet('df_encoded.parquet')
df['neighbourhood group'][0:2500] = df['neighbourhood group'][0:2500].apply(lambda x : 'Manhattan')
df['neighbourhood group'][2500:5000] = df['neighbourhood group'][0:2500].apply(lambda x : 'Brooklyn')
df['neighbourhood group'][5000:7500] = df['neighbourhood group'][0:2500].apply(lambda x : 'Queens')
df['neighbourhood group'][7500:] = df['neighbourhood group'][0:2500].apply(lambda x : 'Bronx')
df['location'] = df['neighbourhood group']
df = df[['price', 'sq. meters', 'description', 'location', 'host name', 'cancellation_policy', 'house_rules', 'text_vector_']]
df = df.reset_index(drop=True)
df

model = SentenceTransformer('all-mpnet-base-v2') #all-MiniLM-L6-v2 #all-mpnet-base-v2

#prepare model #we run it anew in the search function every time, after the initial filtering
# nbrs = NearestNeighbors(n_neighbors=3, algorithm='ball_tree').fit(df['text_vector_'].values.tolist())

def closest_number(x):
    closest_numbers = [25, 40, 45, 55, 60, 70]
    closest_number = closest_numbers[0]
    min_distance = abs(x - closest_number)
    for number in closest_numbers[1:]:
        distance = abs(x - number)
        if distance < min_distance:
            closest_number = number
            min_distance = distance
    return closest_number

def search(df, query):
    product = model.encode(query).tolist()
    # product = df.iloc[0]['text_vector_'] #use one of the products as sample

    nbrs = NearestNeighbors(n_neighbors=3, algorithm='ball_tree').fit(df['text_vector_'].values.tolist())
    distances, indices = nbrs.kneighbors([product]) #input the vector of the reference object

    #print out the description of every recommended product
    df_search = df.iloc[list(indices)[0]].drop(['text_vector_'], axis=1) #.sort_values('avgFeedbackScore', ascending=False)

    return df_search.sort_values('price', ascending=False)

def filter_df(df, column_name, filter_type, filter_value):
    if filter_type == '==':
        df_filtered = df[df[column_name]==filter_value]
    elif filter_type == '>=':
        df_filtered = df[df[column_name]>=filter_value]
    elif filter_type == '<=':
        df_filtered = df[df[column_name]<=filter_value]
    return df_filtered

def predict(history, input1, input2, input3, input4):
    history.append([input1, input2, input3, input4])

    print(history)
    df_location = filter_df(df, 'location', '==', input3)
    df_size = filter_df(df_location, 'sq. meters', '==', input2)
    df_price = filter_df(df_size, 'price', '<=', input1)
    df_result = search(df_price, input4)

    prediction = [
        round(statistics.mean([x[0] for x in history])), #price
        closest_number(statistics.mean([x[1] for x in history])), #square meters
        statistics.mode([x[2] for x in history]) #state
    ]

    return df_result, prediction

with gr.Blocks(theme=gr.themes.Soft(primary_hue='amber', secondary_hue='gray', neutral_hue='amber')) as demo:
    history = gr.Variable(value=[]) #beginning
    gr.Markdown(
    """
    # Airbnb Search Engine
    """
    )
    input1 = gr.Slider(100, 1200, value=700, step_size=100, label="Max Price")
    input2 = gr.Radio([25, 40, 45, 55, 60, 70], multiselect=False, label='square meters', value=45)
    input3 = gr.Radio(['Manhattan', 'Brooklyn', 'Queens', 'Bronx'], multiselect=False, label='State', value='Brooklyn')
    input4 = gr.Textbox(label='Query', value='I want to take a break from work 😴!!!')

    btn = gr.Button(value="Search for a Room")
    output1 = gr.Dataframe()
    output2 = gr.Textbox(label='prediction for the next search')
    # btn.click(greet, inputs='text', outputs=['dataframe'])
    btn.click(predict, [history, input1, input2, input3, input4], [output1, output2])
demo.launch(share=False)