Thanks for your help, however my dataset only has 10k rows so I don't believe this is the problem. If this could help, here is my python script: # Cole ou escreva o código do script aqui: import pandas as pd import numpy as np from scipy.interpolate import Rbf # Import Rbf for radial basis function interpolation import matplotlib.pyplot as plt import seaborn as sns from matplotlib.ticker import FuncFormatter # Drop rows where any of the data is missing or infinite dataset = dataset.replace([np.inf, -np.inf], np.nan).dropna(subset=['Day', 'Depth', 'Temperature']) # Remove outlier temperature values dataset.loc[dataset['Temperature'] > 35, 'Temperature'] = np.nan # Initialize variables for all days in the range of selected years year_min = dataset['Year'].min() year_max = dataset['Year'].max() all_days = pd.date_range(start=f"{year_min}-01-01", end=f"{year_max}-12-31").date day_range = np.arange(1, len(all_days) + 1) depth_range = np.arange(np.floor(dataset['Depth'].min()), np.ceil(dataset['Depth'].max()) + 1) # Initialize the grid with NaNs grid_z = np.full((len(depth_range), len(day_range)), np.nan) # Perform RBF interpolation daily for each depth for day in day_range: day_data = dataset[dataset['Day'] == day] if not day_data.empty: # Filter out rows with NaN values in any key columns day_data = day_data.dropna(subset=['Depth', 'Temperature']) points_y = day_data['Depth'] values = day_data['Temperature'] if len(points_y) > 2: # Ensure there are enough points to interpolate try: rbf_interpolator = Rbf(points_y, values, function='linear', smooth=0.2) valid_depths = np.intersect1d(depth_range, np.round(points_y)) # Interpolate only for available depths grid_z[valid_depths.astype(int) - int(depth_range[0]), day - 1] = rbf_interpolator(valid_depths) except ValueError as e: print(f"Skipping interpolation for day {day} due to insufficient data: {e}") # Create the dataframe for the heatmap heatmap_data = pd.DataFrame(grid_z, index=depth_range, columns=all_days) # Define a function to perform linear interpolation for small gaps in the data def interpolate_small_gaps(data, max_gap=3): for depth_idx in range(data.shape[0]): # Iterate over each depth series = data.iloc[depth_idx] is_na = series.isna() filled = series.copy() for start, group in is_na.groupby((is_na != is_na.shift()).cumsum()): if group.all(): # If the entire group is True (NaNs) if len(group) <= max_gap: # Get the integer indices of the start and end of the gap start_idx = series.index.get_loc(group.index[0]) end_idx = series.index.get_loc(group.index[-1]) # Only interpolate if we have data on both sides of the gap if start_idx > 0 and end_idx < len(series) - 1: interp_range = slice(series.index[start_idx - 1], series.index[end_idx + 1]) filled[interp_range] = series[interp_range].interpolate() data.iloc[depth_idx] = filled return data # Apply the interpolation function to fill small gaps heatmap_data = interpolate_small_gaps(heatmap_data) # Custom function to format the x-axis labels to show only months in Portuguese def custom_formatter(x, pos): index = int(x) # Convert to int to use as index if index < len(all_days): date = all_days[index - 1] # Adjust for zero-based index if date.strftime('%d') == '01': # First day of the month # Mapping English month names to Portuguese month_map = { 'Jan': 'Jan', 'Feb': 'Fev', 'Mar': 'Mar', 'Apr': 'Abr', 'May': 'Mai', 'Jun': 'Jun', 'Jul': 'Jul', 'Aug': 'Ago', 'Sep': 'Set', 'Oct': 'Out', 'Nov': 'Nov', 'Dec': 'Dez' } return month_map[date.strftime('%b')] return '' # Custom function to format the y-axis labels as integers def int_formatter(x, pos): return '%d' % x # Plotting the heatmap plt.figure(figsize=(20, 8)) heatmap = sns.heatmap(heatmap_data, cmap='jet', cbar_kws={'label': 'Temperatura (°C)'}, vmin=0, vmax=30) # Use the custom_formatter function when setting x-tick labels heatmap.set_xticks(np.arange(len(all_days))) heatmap.set_xticklabels([custom_formatter(x, None) for x in range(1, len(all_days) + 1)], rotation=90) # Add the vertical dotted lines and adjust their style based on the day for day, date in enumerate(all_days, start=1): if date.day in [5, 10, 15, 20, 25]: plt.axvline(x=day - 1, color='grey', linestyle=':', linewidth=1) # Thicker line for specified days if date.day == 1: plt.axvline(x=day - 1, color='black', linestyle=':', linewidth=1) # Thicker line for the first day of the month else: plt.axvline(x=day - 1, color='grey', linestyle=':', linewidth=0.5) plt.xlabel('Dia e Mês', fontsize=15) plt.ylabel('Profundidade (m)', fontsize=15) # Apply the custom formatter to the y-axis plt.gca().yaxis.set_major_formatter(FuncFormatter(int_formatter)) plt.tight_layout() plt.show()
