Streamlit_DataDiscovery_Prediction_combined.py

import streamlit as st
import pandas as pd
import numpy as np
import pickle 
from sklearn.ensemble import RandomForestRegressor

st.title("Art Valuation App \U0001F3A8")
st.header('CS Project by Group 2.2')
df = pd.read_csv(r"C:\Users\loren\OneDrive\Documents\5 Semester\CS\Groupe project\cleaned_tow.csv")

st.title('Data Discovery')
# The columns are given new names so that the table looks a little better. 
names_new_columns = {'artist_name':'Artist Name','year_artist_born':'Birth year','age':'Age','length':'Length','width':'Width','title_of_work':'Title','height':'Height'
                     ,'year_of_work':'Creation Year','age_at_work':'Age of Artist at Creation','lower_est_USD':'Lower est. USD','upper_est_USD':'Upper est. USD',
                     'sold_in_USD':'Sales Amount in USD','Material_Category':'Material Category','act_area':'Act. Area','posthumous_combined':'Posthumous Combined'}
df.rename(columns=names_new_columns, inplace= True) #The new names are implimented 

#SLIDER FOR YEAR OF WORK RANGE#
##### we need to convert Creation Year to numeric, because otherwise compares int with string####
st.header('Filter for artworks by year')
df['Creation Year'] = pd.to_numeric(df['Creation Year'], errors='coerce')

# Year range slider code 
min_year = int(df['Creation Year'].min()) if not pd.isnull(df['Creation Year']).all() else 0
max_year = int(df['Creation Year'].max()) if not pd.isnull(df['Creation Year']).all() else 0
selected_years = st.slider("Select a range of years", min_year, max_year, (min_year, max_year))

# filtered df by year of work and display on the app
filtered_df = df[
    (df['Creation Year'] >= selected_years[0]) & 
    (df['Creation Year'] <= selected_years[1])]

st.write("Filtered Artworks by Year:")
st.write(filtered_df)

# Counter for the number of artworks in the selected range
num_artworks = filtered_df.shape[0]
st.markdown(f"Number of Artworks in Selected Range: **{num_artworks}**")

### TEXT INPUT BOX for ARTIST NAME SEARCH ######3
st.header('Search for an artist by name')
artist_name_input = st.text_input("Please enter an artist's name (or part of the name):")
if artist_name_input:
    filtered_df = filtered_df[filtered_df['Artist Name'].str.contains(artist_name_input, case=False)]

# Show search result summary
st.subheader("Results of your Search")
if artist_name_input:
    st.write(f"Showing artworks by artist(s) containing '{artist_name_input}':")
    st.write(filtered_df)
else:
    st.write("Please enter an artist's name (or part of it) to perform a search.")

# Filter for artworks py price with a option of posthumous checkbox
st.header('Filter for artworks by price')
# checkbox to filter posthumous artworks
posthumous_checkbox = st.checkbox("Filter Posthumous Artworks")

# extract Sales Amount in USD column, remove unnecessary caracters, and convert to float
sold_in_usd = df['Sales Amount in USD']

# slider to filter based on sold_in_USD
min_price, max_price = st.slider("Filter by Price Range", 
                                min_value=float(sold_in_usd.min()),
                                max_value=float(sold_in_usd.max()),
                                value=(float(0), float(sold_in_usd.max())))

# filter the dataframe based on the checkbox and slider values
filtered_2_df = df.copy() 

if posthumous_checkbox: # only true is considered posthumous
    filtered_2_df = filtered_2_df[filtered_2_df['Posthumous Combined'].isin([1])]

filtered_2_df['Sales Amount in USD'] = sold_in_usd
filtered_2_df = filtered_2_df[(filtered_2_df['Sales Amount in USD'] >= min_price) & (filtered_2_df['Sales Amount in USD'] <= max_price)]

# Display the dataframe
st.write("Filtered Artworks by Price:")
st.write(filtered_2_df)

# Counter for the number of artworks in the selected range
num_artworks_price = filtered_2_df.shape[0]
st.markdown(f"Number of Artworks in Selected Range: **{num_artworks_price}**")


# Slider for size of the work (area column)
st.header('Filter for artworks by Size')

# Extract this column from the dataframe 
act_area = df['Act. Area']

# Slider to filter based on Act. Area
min_area, max_area = st.slider("Filter by Size Range", 
                                min_value=float(act_area.min()),  #The smallest data item in the dataframe 
                                max_value=float(act_area.max()),  #The biggest data item in the dataframe 
                                value=(float(0), float(act_area.max())))

# Filter the dataframe based on the checkbox and slider values
filtered_3_df = df.copy()

filtered_3_df['Act. Area'] = act_area
filtered_3_df = filtered_3_df[(filtered_3_df['Act. Area'] >= min_area) & (filtered_3_df['Act. Area'] <= max_area)]

# Display of the dataframe with the option of filtering by act. area 
st.write("Filtered Artworks by Size:")
st.write(filtered_3_df)

# Counter for the number of artworks in the selected range
num_artworks_size = filtered_3_df.shape[0]
st.markdown(f"Number of Artworks in Selected Range: **{num_artworks_size}**")


# Multi-checkbox, Filter by Material Category 
st.header('Filter for artworks by Material Category')

# Define the filter options 
# and we group the artworks into ten groups and assign a number to each group so that we can work without the str
filter_options = {
    'Paints and Pigments': 1,
    'Metals and Sculpting Materials': 2,
    'Printmaking and Graphic Arts': 3,
    'Drawing and Writing Tools': 4,
    'Mixed Media and Miscellaneous': 5,
    'Traditional and Specialized Techniques': 6,
    'Photography and Digital Art': 7,
    'Sculpting and Carving': 8,
    'Unconventional and Unique Techniques': 9,
    'Other Specific Techniques or Materials': 10 }

# Checkbox for showing more information if required by the user
# Information about categories thanks to a dictionary 
categories = {
    'Paints and Pigments': ['This category contains artworks made of','oil', 'acrylic', 'spray', 'pastel', 'tempera', 'watercolor', 'watercolour', 'pigment', 'color'],
    'Metals and Sculpting Materials' : ['This category contains artworks made of','bronze', 'gold', 'patinated', 'wrought', 'alabaster', 'polished', 'copper', 'stainless', 'wood', 'cast', 'silver', 'steel', 'aluminum', 'walnut', 'plumbago', 'stone', 'corten', 'metallised', 'lead', 'burnished', 'brass', 'earthenware'],
    'Printmaking and Graphic Arts': ['This category contains artworks made of','cprint', 'offset', 'charcoal', 'screenprint', 'letterpress', 'etching'],
    'Drawing and Writing Tools': ['This category contains artworks made of','ballpoint', 'pen', 'signed', 'crayon', 'pencil', 'conte', 'marker', 'felttip'],
    'Mixed Media and Miscellaneous': ['This category contains artworks made of','embroidery', 'synthetic', 'variable', 'mixed', 'ikb', 'paper', 'painted', 'polymer', 'glazed', 'alkyd', 'resin', 'polyuréthane', 'household', 'digital', 'book', 'fabric', 'gelatin', 'sump', 'vinyl', 'polyester', 'dye', 'offsetlithographin', 'monoprint', 'gypsum', 'printed', 'vegetable', 'archival', 'wool', 'digitally', 'enamel', 'monotype', 'cardboard', 'roll', 'woodcut', 'cement', 'lacquer', 'mirrored', 'oilstick', 'collage', 'pva', 'metal', 'bamboo', 'wires', 'wire', 'sand', 'epson', 'ektacolor', 'unframed', 'peacock', 'plastic', 'flashe', 'water', 'oxidation', 'porcelain', 'molave', 'handknitted', 'embossed', 'handembellished', 'polystone', 'silk', 'vacuumformed', 'cobblestones', 'feathers', 'diasecmounted', 'ukiyoe', 'giclee', 'giclée', 'handpainted', 'homemade', 'cotton', 'duratran', 'indian', 'varnished', 'handcut', 'fibreglass', 'vintage', 'humbrol'],
    'Traditional and Specialized Techniques': ['This category contains artworks made of','silkscreen', 'encaustic', 'terracotta', 'leather', 'screen', 'lambda', 'glass'],
    'Photography and Digital Art': ['This category contains artworks made of','ilfochrome', 'chromogenic', 'ink', 'cibachrome', 'photogram', 'inkjet', 'stencil', 'tar', 'photographic', 'epoxy', 'plaster'],
    'Sculpting and Carving': ['This category contains artworks made of','graphite', 'wax', 'chalk'],
    'Unconventional and Unique Techniques': ['This category contains artworks made of','language', 'led', 'electronic', 'neon', 'kinetic'],
    'Other Specific Techniques or Materials': ['This category contains artworks made of','Other Specific Techniques or Materials: Any material that doesnt fit neatly into the above categories']
}
# Checkbox for showing more information
show_more_info = st.checkbox("Show More Information on the different categories of materials")

# Gives users the option of selecting the categories they want more information on 
if show_more_info:
    st.write("Select Categories for More Information:")
    selected_info_categories = st.multiselect("Select Categories", categories)

    # Display additional information for the selected categories
    for i in selected_info_categories:
        st.write(f"Additional Information for Category {i}:")
        st.write(categories.get(i, "No information available"))

# Multicheckbox for filtering based on categories
selected_categories = st.multiselect("Select Material Category", filter_options.keys()) # Specify keys() so that it doesn't display the values and the user doesn't understand what the number corresponds to.

# Map selected categories to their corresponding values
selected_values = [filter_options[category] for category in selected_categories]

# Filter the DataFrame based on selected values
filtered_4_df = df[df['Material Category'].isin(selected_values)]

# Display the filtered DataFrame based on Material Category 
st.write("Filtered Artworks by Material Category:")
st.write(filtered_4_df)

# Counter for the number of artworks in the selected range
num_artworks_material = filtered_4_df.shape[0]
st.markdown(f"Number of Artworks in Selected Range: **{num_artworks_material}**")


###Section 2 for making predictions 
#Large title to show that users are moving on to another part of the application 
st.title('Artwork Sales Prediction App')

def load_data():
    data = pd.read_csv(r"C:\Users\loren\OneDrive\Documents\5 Semester\CS\Groupe project\My first appprosper_data_app_dev.csv")
    return (data.dropna())

# Calls upon the prediction model
def load_model():
    filename = r"C:\Users\loren\OneDrive\Documents\5 Semester\CS\Groupe project\finalized_artworks_prediction.sav"
    loaded_model = pickle.load(open(filename, "rb"))
    return (loaded_model)

# Load Data and Model 
data = load_data()
model = load_model()

st.write('Enter the following details about your artwork please')
# User inputs of his artwork 
birth_year_input = st.number_input('Birth year', min_value=0, max_value=2023, step=1)
age_input = st.number_input('Age', min_value = 0, max_value = 120, step = 1)
length_input = st.number_input('Length', min_value=0.00, step= 0.001) #min value and steps are float 
width_input = st.number_input('Width', min_value=0.00, step= 0.001)
height_input = st.number_input('Height', min_value=0.00, step=0.001)
creation_year_input = st.number_input('Creation Year', min_value=0, max_value=2023, step=1)
age_at_creation_input = st.number_input('Age of Artist at Creation', min_value=0, max_value=120, step=1)
lower_est_usd_input = st.number_input('Lower est. USD', min_value=0.00, step=0.01)
upper_est_usd_input = st.number_input('Upper est. USD', min_value=0.00, step=0.01)
material_category_input = st.selectbox('Material_Category', filter_options.values()) #Here we must specify that it is the values, otherwise the code will take the keys
act_area_input = length_input*width_input*height_input #The result of multiplying the three is act_area, the user does not need the act. area 
posthumous_combined_input = st.selectbox('Posthumous Combined', [0, 1])
artist_rank_input = st.number_input('Artist rank', min_value = 0 , step = 1)

# Combine user inputs into a DataFrame for prediction
user_input_df = pd.DataFrame({
    'year_artist_born': [birth_year_input],
    'age' : [age_input],
    'length': [length_input],
    'width': [width_input],
    'height': [height_input],
    'year_of_work': [creation_year_input],
    'age_at_work': [age_at_creation_input],
    'lower_est_USD': [lower_est_usd_input],
    'upper_est_USD': [upper_est_usd_input],
    'Material_Category': [material_category_input],
    'act_area' : [act_area_input],
    'posthumous_combined': [posthumous_combined_input],
    'artist_rank' : [artist_rank_input] })

#The input data is run through our prediction model 
input_prediction = model.predict(user_input_df)
st.write(f'The result of the prediction is as follows:', input_prediction)