How to extract reflectance values from points within a many image tiles to build a SVM model that can classify each tile using a fields classes

Question

I would like to run python code in google colab that loops through each tile and extracts the reflectance values from the points within each tile to build a SVM model. Once the model is trained using all the point within each tile the code should be able to classify each tile one by one to create a classified map of my area on interest

`Code which i have so far : 
# Import the necessary libraries.
import ee
!pip install geemap
import geemap

drive.mount('/content/drive')

# Load the shapefile representing the AOI
shapefile_path = '/content/drive/My Drive/CF/_Boarder.shp'
geometry = geemap.shp_to_ee(shapefile_path)
# Define the start and end dates for your image collection.
start_date = '2023-11-14'
end_date = '2023-11-27'

# Define the filter for cloud cover.
cloud_filter = ee.Filter.lt('CLOUD_COVERAGE_ASSESSMENT', 50)

# Load the Sentinel 2 imagery collection.
s2_collection = (ee.ImageCollection('COPERNICUS/S2_SR')
                  .filterDate(start_date, end_date)
                  .filterBounds(geometry)
                  .filter(cloud_filter)
                  .sort('system:time_start', True))

# Define the path to the shapefile
Sample_points_shapefile_path = '/content/drive/My Drive/CF/Raw_Data_23022024.shp'

# Load the shapefile as a GeoDataFrame
gdf = geemap.shp_to_gdf(Sample_points_shapefile_path)

# Convert the GeoDataFrame to an ee.FeatureCollection
sample_FC = geemap.gdf_to_ee(gdf)

# Visualize the first few rows of the GeoDataFrame
print(gdf.head())

   Random_ID   IAP_Yes_or    Fynbos_typ Alien_taxa  \ 
0        774          Yes    Not fynbos        Gum   
1        554          Yes  Scrub Fynbos       Pine   
2        135          No  Scrub Fynbos       None   
3        729          Yes  Scrub Fynbos       Pine   
4        771          Yes    Not fynbos       Pine 
            LULC__Fynb   Latitude  Longitude                    geometry  
0                  Gum -34.595878  19.568680  POINT (19.56868 -34.59588)  
1                 Pine -34.587340  19.552233  POINT (19.55223 -34.58734)  
2  High Density Fynbos -34.587086  19.548819  POINT (19.54882 -34.58709)  
3                 Pine -34.585369  19.528607  POINT (19.52861 -34.58537)  
4                 Pine -34.593895  19.519623  POINT (19.51962 -34.59389)

**I want to use the field LULC__Fynb as classes for the classification **

# Get a list of unique Sentinel-2 tiles intersecting with your AOI
tile_collection = s2_collection.aggregate_array('MGRS_TILE').distinct().getInfo()

# Print the list of tiles
print("Sentinel-2 Tiles intersecting with the AOI:")
print(tile_collection)
Sentinel-2 Tiles intersecting with the AOI:
['34JCL', '34JBL', '33JYF', '34HCK', '34HBK', '33HYE', '34HBJ', '33HYD', '34HBH', '33HYC', '35HMD', '35HLD', '35HKD', '34HGJ', '35HMC', '34HFJ', '35HLC', '35HKC', '34HGH', '34HFH', '34HEJ', '34HDJ', '34HEH', '34HCG', '35HND', '35HNC', '34HDH', '34HEG', '34HDG', '34HCJ', '34HCH', '34HBG']






**Somthing along the lines of:**


# Define function to extract reflectance values from points within each tile
def extract_values(tile, geometry, collection):
    # Filter collection by tile
    tile_collection = collection.filter(ee.Filter.eq('MGRS_TILE', tile))
    
    # Function to extract reflectance values for a single image
    def extract_point_values(feature):
        image = ee.Image(tile_collection.first())  # Get the first image in the collection
        return feature.set('reflectance', image.reduceRegion(
            reducer=ee.Reducer.first(),
            geometry=feature.geometry(),
            scale=10
        ))
    
    return geometry.map(extract_point_values)

# Loop through each tile
for tile in tile_collection:
    print("Processing tile:", tile)
    
    # Extract reflectance values for points within the tile
    points_with_reflectance = extract_values(tile, sample_FC, s2_collection)
    
    # Convert feature collection to numpy array
    points_array = fc_to_array(points_with_reflectance)
    
    # Extract labels (LULC__Fynb)
    labels = gdf['LULC__Fynb']
    
    # Identify missing labels
    missing_labels = set(labels.unique()) - set(label_mapping.keys())
    
    # Update label mapping to include missing labels
    for label in missing_labels:
        label_mapping[label] = len(label_mapping)
    
    # Convert labels to numerical values
    labels_numeric = labels.map(lambda x: label_mapping[x])
    



`