durhamtrees.py

# -*- coding: utf-8 -*-
"""DurhamTrees

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1czig7JIbqTKp9wNUIRcdMEDF3pFgtxKv
"""

# -*- coding: utf-8 -*-
"""DurhamTrees
Automatically generated by Colaboratory.
Original file is located at
    https://colab.research.google.com/drive/1czig7JIbqTKp9wNUIRcdMEDF3pFgtxKv
"""
import pyarrow.parquet as pq
import pandas as pd
import geopandas as gpd
from datasets import (
    GeneratorBasedBuilder, Version, DownloadManager, SplitGenerator, Split,
    Features, Value, BuilderConfig, DatasetInfo
)
import matplotlib.pyplot as plt
import seaborn as sns
import csv
import json
from shapely.geometry import Point
import base64
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
import io
# URL definitions
_URLS = {
    "first_domain1": {
        "csv_file": "https://drive.google.com/uc?export=download&id=18HmgMbtbntWsvAySoZr4nV1KNu-i7GCy",
        "geojson_file": "https://drive.google.com/uc?export=download&id=1cbn7EY7RofXN7c6Ph0eIGFIZowPZ5lKE",
        "parquet_file": "https://drive.google.com/uc?export=download&id=1RNDLJLoSSV9RJptVyfWFhPra0nh-i_CN",
    },
    "first_domain2": {
        "csv_file2": "https://drive.google.com/uc?export=download&id=1RVdaI5dSTPStjhOHO40ypDv2cAQZpi_Y",
    },
}

# Combined Dataset Class
class DurhamTrees(GeneratorBasedBuilder):
    VERSION = Version("1.0.0")

    class MyConfig(BuilderConfig):
        def __init__(self, **kwargs):
            super().__init__(**kwargs)

    BUILDER_CONFIGS = [
        MyConfig(name="class1_domain1", description="this is combined of csv and geojson"),
        MyConfig(name="class2_domain1", description="this is csv file"),
    ]

    def _info(self):
        return DatasetInfo(
            description="This dataset combines information from both classes, with additional processing for csv_file2.",
            features=Features({
                "feature1_from_class1": Value("string"),
                "geometry":Value("string"),
                "OBJECTID": Value("int64"),
                "X": Value("float64"),
                "Y": Value("float64"),
                "image": Value("binary"),
                "label": Value("int64"),
                "feature1_from_class2": Value("string"),
                "streetaddress": Value("string"),
                "city": Value("string"),
                "facilityid": Value("int64"),
                "present": Value("string"),
                "genus": Value("string"),
                "species": Value("string"),
                "commonname": Value("string"),
                "diameterin": Value("float64"),
                "condition": Value("string"),
                "neighborhood": Value("string"),
                "program": Value("string"),
                "plantingw": Value("string"),
                "plantingcond": Value("string"),
                "underpwerlins": Value("string"),
                "GlobalID": Value("string"),
                "created_user": Value("string"),
                "last_edited_user": Value("string"),
                "isoprene": Value("float64"),
                "monoterpene": Value("float64"),
                "monoterpene_class2": Value("float64"),
                "vocs": Value("float64"),
                "coremoved_ozperyr": Value("float64"),
                "coremoved_dolperyr": Value("float64"),
                "o3removed_ozperyr": Value("float64"),
                "o3removed_dolperyr": Value("float64"),
                "no2removed_ozperyr": Value("float64"),
                "no2removed_dolperyr": Value("float64"),
                "so2removed_ozperyr": Value("float64"),
                "so2removed_dolperyr": Value("float64"),
                "pm10removed_ozperyr": Value("float64"),
                "pm10removed_dolperyr": Value("float64"),
                "pm25removed_ozperyr": Value("float64"),
                "o2production_lbperyr": Value("float64"),
                "replacevalue_dol": Value("float64"),
                "carbonstorage_lb": Value("float64"),
                "carbonstorage_dol": Value("float64"),
                "grosscarseq_lbperyr": Value("float64"),
                "grosscarseq_dolperyr": Value("float64"),
                "avoidrunoff_ft2peryr": Value("float64"),
                "avoidrunoff_dol2peryr": Value("float64"),
                "polremoved_ozperyr": Value("float64"),
                "polremoved_dolperyr": Value("float64"),
                "totannbenefits_dolperyr": Value("float64"),
                "leafarea_sqft": Value("float64"),
                "potevapotran_cuftperyr": Value("float64"),
                "evaporation_cuftperyr": Value("float64"),
                "transpiration_cuftperyr": Value("float64"),
                "h2ointercept_cuftperyr": Value("float64"),
                "carbonavoid_lbperyr": Value("float64"),
                "carbonavoid_dolperyr": Value("float64"),
                "heating_mbtuperyr": Value("float64"),
                "heating_dolperyrmbtu": Value("float64"),
                "heating_kwhperyr": Value("float64"),
                "heating_dolperyrmwh": Value("float64"),
                "cooling_kwhperyr": Value("float64"),
                "cooling_dolperyr": Value("float64"),
                "totalenerg_dolperyr": Value("float64"),
            }),
            supervised_keys=("image", "label"),
            homepage="https://github.com/AuraMa111?tab=repositories",
            citation="Citation for the combined dataset",
        )

    def _split_generators(self, dl_manager):
          downloaded_files = dl_manager.download_and_extract(_URLS)

          return [
              SplitGenerator(
                  name=Split.TRAIN,
                  gen_kwargs={
                      "class1_data_file": downloaded_files["first_domain1"]["csv_file"],
                      "class1_geojson_file": downloaded_files["first_domain1"]["geojson_file"],
                      "class2_data_file": downloaded_files["first_domain2"]["csv_file2"],
                      "parquet_file": downloaded_files["first_domain1"]["parquet_file"],
                      "split": Split.TRAIN,
                  },
              ),
          ]




    def _generate_examples(self, class1_data_file, class1_geojson_file, class2_data_file, parquet_file, split):
        class1_examples = list(self._generate_examples_from_class1(class1_data_file, class1_geojson_file))
        class2_examples = list(self._generate_examples_from_class2(class2_data_file))

        # Load Parquet file
        parquet_data = pq.read_table(parquet_file).to_pandas()
        class1_examples += list(self._generate_examples_from_parquet(parquet_data))

        examples = class1_examples + class2_examples
        df = pd.DataFrame(examples)

        for id_, example in enumerate(examples):
            if not isinstance(example, dict):
                # Convert the example to a dictionary if it's not
                example = {"example": example}
            yield id_, example

    def _generate_examples_from_class1(self, csv_filepath, geojson_filepath):
          columns_to_extract = ["OBJECTID", "X", "Y"]  # Remove "geometry" from columns_to_extract
          csv_data = pd.read_csv(csv_filepath)

          with open(geojson_filepath, 'r') as file:
              geojson_dict = json.load(file)
          gdf = gpd.GeoDataFrame.from_features(geojson_dict['features'], crs="EPSG:4326")  # Specify the CRS if known
          merged_data = gdf.merge(csv_data, on='OBJECTID')
          final_data = merged_data[columns_to_extract + ['geometry']]  # Include 'geometry' in the final_data
          for id_, row in final_data.iterrows():
              example = row.to_dict()
              yield id_, example





    def _generate_examples_from_class2(self, csv_filepath2):
        csv_data2 = pd.read_csv(csv_filepath2)


        columns_to_extract = [
            "streetaddress", "city", "facilityid", "present", "genus", "species",
            "commonname", "diameterin", "condition",  "neighborhood", "program", "plantingw",
            "plantingcond", "underpwerlins", "GlobalID", "created_user", "last_edited_user", "isoprene", "monoterpene",
             "monoterpene", "vocs", "coremoved_ozperyr", "coremoved_dolperyr",
            "o3removed_ozperyr", "o3removed_dolperyr", "no2removed_ozperyr", "no2removed_dolperyr",
            "so2removed_ozperyr", "so2removed_dolperyr", "pm10removed_ozperyr", "pm10removed_dolperyr",
            "pm25removed_ozperyr", "o2production_lbperyr", "replacevalue_dol", "carbonstorage_lb",
            "carbonstorage_dol", "grosscarseq_lbperyr", "grosscarseq_dolperyr", "polremoved_ozperyr", "polremoved_dolperyr",
            "totannbenefits_dolperyr", "leafarea_sqft", "potevapotran_cuftperyr", "evaporation_cuftperyr",
            "transpiration_cuftperyr", "h2ointercept_cuftperyr",
             "carbonavoid_lbperyr", "carbonavoid_dolperyr", "heating_mbtuperyr",
            "heating_dolperyrmbtu", "heating_kwhperyr", "heating_dolperyrmwh", "cooling_kwhperyr",
            "cooling_dolperyr", "totalenerg_dolperyr",
        ]

        final_data = csv_data2[columns_to_extract]
        for id_, row in final_data.iterrows():
            example = row.to_dict()
            non_empty_example = {key: value for key, value in example.items() if pd.notna(value)}
            yield id_, example


    def _generate_examples_from_parquet(self, parquet_data):
        for id_, row in parquet_data.iterrows():
            # Check if the "image" column is present and not empty
            if "image" in row and "bytes" in row["image"]:
                # Decode the base64-encoded image bytes
                image_data = base64.b64decode(row["image"]["bytes"])
                example = {"image": image_data, "label": row["label"]}

                # Display the image
                image_bytes = example.get('image', None)
                if image_bytes:
                    img = mpimg.imread(io.BytesIO(image_bytes), format='PNG')  # Use 'PNG' instead of 'JPG'
                    plt.imshow(img)
                    plt.show()

                yield id_, example
            else:
                print(f"Skipping example {id_} as it has missing or invalid image data")



    def _correlation_analysis(self, df):
        correlation_matrix = df.corr()
        sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', linewidths=.5)
        plt.title("Correlation Analysis")
        plt.show()






# Create an instance of the DurhamTrees class
durham_trees_dataset = DurhamTrees(name='class1_domain1')

# Build the dataset
durham_trees_dataset.download_and_prepare()

# Access the dataset
dataset = durham_trees_dataset.as_dataset()

# Iterate through the dataset and display images
for example in dataset['train']:
    if "image" in example and example["image"] is not None and "bytes" in example["image"]:
        # Display the image
        image_data = base64.b64decode(example["image"]["bytes"])
        img = mpimg.imread(io.BytesIO(image_data), format='PNG')
        plt.imshow(img)
        plt.show()

# Create an instance of the DurhamTrees class for another configuration
durham_trees_dataset_another = DurhamTrees(name='class2_domain1')

# Build the dataset for the new instance
durham_trees_dataset_another.download_and_prepare()

# Access the dataset for the new instance
dataset_another = durham_trees_dataset_another.as_dataset()

# Iterate through the dataset for the new instance and display images
for example in dataset_another['train']:
    if "image" in example and example["image"] is not None and "bytes" in example["image"]:
        # Display the image
        image_data = base64.b64decode(example["image"]["bytes"])
        img = mpimg.imread(io.BytesIO(image_data), format='PNG')
        plt.imshow(img)
        plt.show()