import numpy as np import pandas as pd from numba import jit import math import json import os import sys from specklepy.api.client import SpeckleClient from specklepy.api.credentials import get_default_account, get_local_accounts from specklepy.transports.server import ServerTransport from specklepy.api import operations from specklepy.objects.geometry import Polyline, Point from specklepy.objects import Base from specklepy.api import operations, models from specklepy.transports.server import ServerTransport import time from functools import wraps import sys sys.path.append("speckleUtils") import speckle_utils #https://serjd-syncspeckle2notion.hf.space/webhooks/update_streams #https://serjd-RECODE_HF_tripGeneration.hf.space/webhooks/update_streams #serJD/RECODE_HF_tripGeneration # https://huggingface.co/spaces/serJD/RECODE_HF_tripGeneration # !!! lots of hard coded values in computeTrips !!! # UTILS def reconstruct_dataframe(alpha_low, alpha_med, alpha_high, original_df): # Define the mapping from original values to new alpha parameters value_to_alpha = { 0.00191: alpha_low, 0.00767: alpha_high, 0.0038: alpha_med } # Check if each value is present at least once in the DataFrame for original_value in value_to_alpha.keys(): if not (original_df == original_value).any().any(): raise ValueError(f"Value {original_value} not found in the input DataFrame.") # Create a new DataFrame based on the original one new_df = original_df.copy() # Apply the mapping to each element in the DataFrame for original_value, new_value in value_to_alpha.items(): new_df = new_df.replace(original_value, new_value) return new_df def preprocess_dataFrame(df, headerRow_idx=0, numRowsStart_idx = None, numRowsEnd_idx=None, numColsStart_idx=None, numColsEnd_idx=None, rowNames_idx=None): df.columns = df.iloc[headerRow_idx] #Set the header if rowNames_idx is not None: df.index = df.iloc[:, rowNames_idx] #Set the row names df = df.iloc[numRowsStart_idx : numRowsEnd_idx, numColsStart_idx:numColsEnd_idx] #Slice the dataset to numerical data return df def timeit(f): def timed(*args, **kw): ts = time.time() result = f(*args, **kw) te = time.time() print ('func:%r args:[%r, %r] took: %2.4f sec' % \ (f.__name__, te-ts)) #(f.__name__, args, kw, te-ts)) return result return timed def timing_decorator(func): @wraps(func) def wrapper(*args, **kwargs): start_time = time.time() result = func(*args, **kwargs) end_time = time.time() duration = end_time - start_time timestamp = time.strftime('%Y-%m-%d %H:%M:%S', time.localtime(end_time)) print(f"{func.__name__} took {duration:.4f} seconds. Finished at {timestamp}") return result return wrapper # Function to compare two dataframes after converting and rounding def compare_dataframes(df1, df2, decimals=8): # Function to convert DataFrame columns to float and then round def convert_and_round_dataframe(df, decimals): # Convert all columns to float df_float = df.astype(float) # Round to the specified number of decimals return df_float.round(decimals) rounded_df1 = convert_and_round_dataframe(df1, decimals) rounded_df2 = convert_and_round_dataframe(df2, decimals) are_equal = rounded_df1.equals(rounded_df2) print("Both methods are equal:", are_equal) print("Numba shape:", df2.shape) print("Original shape:", df1.shape) print("======== ORIGINAL OUTPUT (first item in output list, head() for the first 5 columns)") print(df1.iloc[0:5].head(2)) print("======== New method OUTPUT (first item in output list, head() for the first 5 columns)") print(df2.iloc[0:5].head(2)) def align_dataframes(df1, df2, key): """ Align two dataframes based on a common key, ensuring that both dataframes have only the rows with matching keys. Parameters: - df1: First dataframe. - df2: Second dataframe. - key: Column name to align dataframes on. Returns: - df1_aligned, df2_aligned: Tuple of aligned dataframes. """ common_ids = df1.index.intersection(df2[key]) df1_aligned = df1.loc[common_ids] df2_aligned = df2[df2[key].isin(common_ids)].set_index(key, drop=False) return df1_aligned, df2_aligned #================================================================================================== def attraction_proNode(df_attraction_num, df_lu, df_lu_anName=None, printSteps=False): #lu_proNode df_lu_proNode = df_lu.reset_index()[df_lu_anName] if printSteps: print(df_lu_proNode.shape) df_lu_proNode.head(50) #attraction_proNode if printSteps: print("df_attraction_num:", df_attraction_num.shape) print("df_lu_proNode:", df_lu_proNode.shape) df_attraction_proNode = df_attraction_num.mul(df_lu_proNode, axis=0) if printSteps: print("df_attraction_proNode:", df_attraction_proNode.shape) df_attraction_proNode.head(100) # Sum the values of each column df_attraction_proNode_sum = pd.DataFrame(df_attraction_proNode.sum(), columns=['Sum']) if printSteps: print("df_attraction_proNode_sum:", df_attraction_proNode_sum.shape) df_attraction_proNode_sum.head(100) return df_attraction_proNode_sum #Non vectorized iterative function def attraction_proNode_full_iter(df_attraction_num, df_lu_num, printSteps=False): # Initialize an empty DataFrame df_attraction_proNode_sum_total = pd.DataFrame() for column_name, column_data in df_lu_num.items(): df_attraction_proNode_sum = attraction_proNode(df_attraction_num, df_lu_num, df_lu_anName=column_name) # Concatenate DataFrames along columns df_attraction_proNode_sum_total = pd.concat([df_attraction_proNode_sum_total, df_attraction_proNode_sum], axis=1) # Rename columns in df_distBasedAttr_step2 with the same column names as in df_distributionMatrix_step1 df_attraction_proNode_sum_total.columns = df_lu_num.columns return df_attraction_proNode_sum_total # PRODUCTION ================================================ def production_proNode(df_lu, df_sqmProPerson, df_tripRate, df_production_num, df_production_transposed, printSteps=False, df_lu_anName=None): #lu_proNode - reset index df_lu_proNode = df_lu.reset_index()[df_lu_anName] if printSteps: print(df_lu_proNode.shape) df_lu_proNode.head(50) #Get the person count - Divide corresponding values of one DataFrame by another df_personCount = df_lu_proNode.div(df_sqmProPerson) if printSteps: print(df_personCount.shape) print(df_personCount) # Ensure the index is unique in df_personCount df_personCount = df_personCount.reset_index(drop=True) df_production_transposed = df_production_transposed.reset_index(drop=True) if printSteps: df_production_transposed.head() if printSteps: df_personCount.head() df_tripRate.head() #Calculate trip production pro node df_production_proNode = df_production_transposed df_production_proNode = df_production_proNode.mul(df_personCount, axis=0) df_production_proNode = df_production_proNode.T df_production_proNode = df_production_proNode.mul(df_tripRate, axis=0) #Total trips df_production_proNode_rowSum = df_production_proNode.sum(axis=1) df_total_trips = df_production_proNode_rowSum #if printSteps: #df_total_trips.head(50) return df_total_trips #Non vectorized iterative function def production_proNode_total(df_lu, df_sqmProPerson, df_tripRate, df_production_num, df_production_transposed, df_lu_num, printSteps=False): # Initialize an empty DataFrame df_total_trips_allNodes = pd.DataFrame() for column_name, column_data in df_lu_num.items(): df_total_trips_proNode = production_proNode(df_lu_num, df_sqmProPerson, df_tripRate, df_production_num, df_production_transposed, printSteps=False, df_lu_anName=column_name) # Concatenate DataFrames along columns df_total_trips_allNodes = pd.concat([df_total_trips_allNodes, df_total_trips_proNode], axis=1) # Rename columns in df_distBasedAttr_step2 with the same column names as in df_distributionMatrix_step1 df_total_trips_allNodes.columns = df_lu_num.columns return df_total_trips_allNodes #df_total_trips_allNodes = production_proNode_total(df_lu, df_sqmProPerson, df_tripRate, df_production_num, df_production_transposed, df_lu_num, printSteps=False) #df_total_trips_allNodes.head(50) #================================================================================================== #STEP 1 def step_1(df_distributionMatrix, df_total_trips_allNodes): l = [] #counter=0 for column_name, column_data in df_total_trips_allNodes.items(): df_distributionMatrix_step1_proNode = df_distributionMatrix.mul(column_data, axis = 0) l.append(df_distributionMatrix_step1_proNode) return l #STEP 2 def step_2_vectorized(df_distMatrix, df_alphas): # Convert df_distMatrix to a 2D array: Shape (1464, 1464) distMatrix_array = df_distMatrix.values # Convert df_alphas to a 1D array: Shape (26,) alphas_array = df_alphas.values # Initialize an empty array to store results: Shape (1464, 1464, 26) result_3d = np.zeros((distMatrix_array.shape[0], distMatrix_array.shape[1], len(alphas_array))) # Loop over alphas and perform element-wise multiplication followed by exponential function for i in range(len(alphas_array)): result_3d[:, :, i] = np.exp(-distMatrix_array * alphas_array[i]) # Construct the final list of DataFrames final_list = [pd.DataFrame(result_3d[i, :, :], columns=df_alphas.index, index=df_distMatrix.index) for i in range(result_3d.shape[0])] return final_list # Step 3 @jit(nopython=True) def multiply_and_sum(arr, attraction_arr): # Element-wise multiplication multiplied_arr = arr * attraction_arr # Sum the values of each column summed_arr = multiplied_arr.sum(axis=0) return multiplied_arr, summed_arr def step_3_numba(df_attraction_proNode_sum_total, df_step_2): # Convert df_attraction_proNode_sum_total to a NumPy array and transpose it attraction_array = df_attraction_proNode_sum_total.values.T.astype(np.float64) # Ensure float64 dtype multiplied_results = [] summed_results = [] for df in df_step_2: # Convert DataFrame to NumPy array with float64 dtype df_array = df.values.astype(np.float64) # Perform element-wise multiplication and summing multiplied_arr, summed_arr = multiply_and_sum(df_array, attraction_array) # Convert results back to DataFrames df_multiplied = pd.DataFrame(multiplied_arr, columns=df.columns, index=df.index) # Reshape summed_arr to have shape (26,1) and then convert to DataFrame df_summed = pd.DataFrame(summed_arr.reshape(-1, 1), index=df.columns, columns=['Sum']) multiplied_results.append(df_multiplied) summed_results.append(df_summed) return multiplied_results, summed_results # step 4 @jit(nopython=True) def divide_and_sum(arr, divisor_arr): # Ensure divisor_arr is broadcastable to arr's shape divisor_arr_expanded = divisor_arr.reshape((divisor_arr.shape[0], 1, divisor_arr.shape[1])) # Initialize arrays to store results divided_result = np.zeros_like(arr) summed_result = np.zeros((arr.shape[0], arr.shape[2])) for i in range(arr.shape[0]): for j in range(arr.shape[1]): for k in range(arr.shape[2]): if divisor_arr_expanded[i, 0, k] != 0: divided_result[i, j, k] = arr[i, j, k] / divisor_arr_expanded[i, 0, k] summed_result[i, k] += divided_result[i, j, k] return divided_result, summed_result def step_4_numba(distAndAreaBasedAttr_step3, distAndAreaBasedAttr_step3_sum): # Convert lists of DataFrames to 3D arrays with dtype float64 array_step3 = np.array([df.values for df in distAndAreaBasedAttr_step3]).astype(np.float64) array_step3_sum = np.array([df.values for df in distAndAreaBasedAttr_step3_sum]).astype(np.float64) # Perform division and summation using Numba divided_result, summed_result = divide_and_sum(array_step3, array_step3_sum) # Convert results back to lists of DataFrames df_distAndAreaBasedAttr_step4 = [pd.DataFrame(divided_result[i], columns=distAndAreaBasedAttr_step3[0].columns, index=distAndAreaBasedAttr_step3[0].index) for i in range(divided_result.shape[0])] # Correct the creation of the summed DataFrame to avoid setting the 'Sum' index df_distAndAreaBasedAttr_step4_sum = [pd.DataFrame(summed_result[i]).T.set_axis(['Sum'], axis='index').set_axis(distAndAreaBasedAttr_step3[0].columns, axis='columns') for i in range(summed_result.shape[0])] return df_distAndAreaBasedAttr_step4, df_distAndAreaBasedAttr_step4_sum # step 5 @jit(nopython=True) def tripsPerArctivity_numba(matrix, attrs): rows, cols = attrs.shape[0], matrix.shape[0] # 1464, 26 result = np.zeros((cols, rows), dtype=np.float64) # Prepare result matrix (26, 1464) for i in range(rows): # Iterate over each area for j in range(cols): # Iterate over each land use category sum_val = 0.0 for k in range(cols): # Iterate over each element in the distribution matrix row sum_val += matrix[j, k] * attrs[i, k] result[j, i] = sum_val return result def step_5_numba(distributionMatrix_step1, distAndAreaBasedAttr_step4): sums = [] count = 0 total_count = len(distributionMatrix_step1) for df_distributionMatrix_step1, df_distAndAreaBasedAttr_step4 in zip(distributionMatrix_step1, distAndAreaBasedAttr_step4): # Convert DataFrames to NumPy arrays with dtype float64 matrix = df_distributionMatrix_step1.values.astype(np.float64) attrs = df_distAndAreaBasedAttr_step4.values.astype(np.float64) result = tripsPerArctivity_numba(matrix, attrs) df_result = pd.DataFrame(result, index=df_distributionMatrix_step1.columns, columns=df_distAndAreaBasedAttr_step4.index) sums.append(df_result) count += 1 #print(f"Iteration {count} out of {total_count} is finished.") #print("---------") return sums # step 6&7 def step_6_7_vectorized(df_trips_proNode_proActivity_total): # Convert each DataFrame to a NumPy array and stack them to form a 3D array array_3d = np.array([df.values for df in df_trips_proNode_proActivity_total]) # Sum across the middle axis (columns of each DataFrame) summed_array = array_3d.sum(axis=1) # Convert the summed array back to a DataFrame final_matrix = pd.DataFrame(summed_array, index=df_trips_proNode_proActivity_total[0].columns, columns=df_trips_proNode_proActivity_total[0].columns) return final_matrix # step 8 def adjTripRate_adjFactor(tripMatrix,df_total_trips_allNodes_sumPerson, targetRate=1, factor=1 ): df_tripMatrix_total_sum = tripMatrix.sum().sum() df_total_trips_allNodes_sumPerson_total = df_total_trips_allNodes_sumPerson.sum() # scale to target trip rate tripRateBeforeAdjustment = df_tripMatrix_total_sum/df_total_trips_allNodes_sumPerson_total print("tripRateBeforeAdjustment",tripRateBeforeAdjustment) adjustmentRate = targetRate/tripRateBeforeAdjustment print("adjustmentRate",adjustmentRate) # scale by ... scale factor (outdated, was hardcoded ) df_tripMatrix_adjusted = tripMatrix * adjustmentRate #df_tripMatrix_adjusted_scaled = df_tripMatrix_adjusted.div(factor) return df_tripMatrix_adjusted, df_tripMatrix_adjusted # df_tripMatrix_adjusted_scaled # Uniform Matrix def decay(d, alpha): return math.exp(d * alpha * -1) def distanceDecay(df, alpha): return df.applymap(lambda x: decay(x, alpha)) def matrix_reduce_add(df): return df[df != sys.float_info.max].sum().sum() def replace_maxValue(df): return df.replace(sys.float_info.max, 0) #Trip gen matrix is used to scale the distance matrix def getUniformMatrix(distanceMatrix, tripGenMatrix, alpha): distanceMatrix_withDecay = distanceDecay(distanceMatrix, alpha) distanceMatrix_sum = matrix_reduce_add(distanceMatrix_withDecay) tripGenMatrix_sum = matrix_reduce_add(tripGenMatrix) ratio = distanceMatrix_sum / tripGenMatrix_sum uniformMatrix = distanceMatrix_withDecay.div(ratio) return replace_maxValue(uniformMatrix) #================================================================================================== #Modal Split functions def computeModalShare(trip_matrix, dist_matrices, alpha, f_values=None): """ Process matrices or DataFrames with exponentiation and normalization. Args: trip_matrix (np.ndarray or pd.DataFrame): The trip matrix. dist_matrices (dict of np.ndarray or pd.DataFrame): Dictionary of distance matrices. alpha (float): The alpha coefficient. f_values (dict of float, optional): Dictionary of f coefficients for each matrix. If None, defaults to 0 for each matrix. Returns: dict: Normalized matrices. """ # Default f_values to 0 for each key in dist_matrices if not provided if not f_values: f_values = {key: 0 for key in dist_matrices.keys()} exp_matrices = {} for key, matrix in dist_matrices.items(): f = f_values.get(key, 0) # Convert DataFrame to numpy array if needed if isinstance(matrix, pd.DataFrame): matrix = matrix.values exp_matrix = np.exp(-1 * (matrix * alpha + f)) exp_matrices[key] = exp_matrix # Calculate the sum of all exponentials sum_exp = sum(exp_matrices.values()) # Normalize each matrix & multiply by trip matrix and update the matrices normalized_matrices = {key: (exp_matrix / sum_exp) * trip_matrix for key, exp_matrix in exp_matrices.items()} return normalized_matrices def redistributeModalShares(dist_matrices, trip_matrices, redistribution_rules, threshold=0.5): """ Redistribute trips among mobility networks based on given redistribution rules and when travel times are within a specified threshold. Args: dist_matrices (dict): Dictionary of distance matrices (travel times) for different mobility networks, keyed by identifier. trip_matrices (dict): Dictionary of matrices representing the number of trips for each mobility network, keyed by identifier. redistribution_rules (list): List of redistribution rules with "from" and "to" network identifiers. threshold (float): The threshold for considering travel times as similar. Returns: dict: Updated dictionary of trip matrices with transferred trips. """ # Verify that all specified matrices exist in the input dictionaries for rule in redistribution_rules: if rule["from"] not in dist_matrices or rule["from"] not in trip_matrices: raise ValueError(f"Matrix ID {rule['from']} not found in the inputs.") for to_id in rule["to"]: if to_id not in dist_matrices or to_id not in trip_matrices: raise ValueError(f"Matrix ID {to_id} not found in the inputs.") # Copy the trip_matrices to avoid modifying the input directly updated_trip_matrices = {k: v.copy() for k, v in trip_matrices.items()} # Redistribute trips based on the rules and the threshold for rule in redistribution_rules: from_matrix_id = rule["from"] from_matrix_trips = updated_trip_matrices[from_matrix_id] from_matrix_dist = dist_matrices[from_matrix_id] for to_matrix_id in rule["to"]: to_matrix_dist = dist_matrices[to_matrix_id] # Create a boolean array where the absolute difference in travel times is less than or equal to the threshold similar_travel_time = np.abs(from_matrix_dist - to_matrix_dist) <= threshold # Find the indices where there are trips to transfer under the new condition indices_to_transfer = similar_travel_time & (from_matrix_trips > 0) # Transfer trips where the condition is True updated_trip_matrices[to_matrix_id][indices_to_transfer] += from_matrix_trips[indices_to_transfer] # Zero out the transferred trips in the from_matrix from_matrix_trips[indices_to_transfer] = 0 # Return the updated trip matrices dictionary return updated_trip_matrices def computeDistanceBrackets(trip_matrices, metric_dist_matrices, dist_brackets=[800, 2400, 4800]): # Transform the keys of metric_dist_matrices to match with trip_matrices transformed_metric_keys = {key.replace("metric_matrix", "distance_matrix")+"_noEntr": matrix for key, matrix in metric_dist_matrices.items()} # Initialize dictionary to store aggregated trips per distance bracket bracket_totals = {bracket: 0 for bracket in dist_brackets} # Iterate over each pair of trip matrix and distance matrix for key, trip_matrix in trip_matrices.items(): # Find the corresponding distance matrix dist_matrix = transformed_metric_keys.get(key) if dist_matrix is None: print("no matrxi found") continue # Skip if no corresponding distance matrix found # Calculate trips for each distance bracket for i, bracket in enumerate(dist_brackets): if i == 0: # For the first bracket, count trips with distance <= bracket bracket_totals[bracket] += (trip_matrix[dist_matrix <= bracket]).sum().sum() else: # For subsequent brackets, count trips within the bracket range prev_bracket = dist_brackets[i - 1] bracket_totals[bracket] += (trip_matrix[(dist_matrix > prev_bracket) & (dist_matrix <= bracket)]).sum().sum() brackets_sum = sum(bracket_totals.values()) brackets_rel = {str(bracket): round(total / brackets_sum, 3) for bracket, total in bracket_totals.items()} return brackets_rel def computeTripStats(trip_matrices, distance_matrices, metric_dist_matrices, pop): # Transform the keys of metric_dist_matrices to match with trip_matrices transformed_metric_keys = {key.replace("metric_matrix", "distance_matrix")+"_noEntr": matrix for key, matrix in metric_dist_matrices.items()} trips = 0 totalTravelDistance = 0 totalTravelTime = 0 # Iterate over each pair of trip matrix and distance matrix for key, trip_matrix in trip_matrices.items(): # Find the corresponding distance matrix metric_dist_matrix = transformed_metric_keys.get(key) dist_matrix = distance_matrices.get(key) if metric_dist_matrix is None: print("no matrxi found") continue # Skip if no corresponding distance matrix found # compute totalTravelTime += (dist_matrix*trip_matrix).sum().sum() trips += trip_matrix.sum().sum() totalTravelDistance += (metric_dist_matrix*trip_matrix).sum().sum() MeanTripDistance = totalTravelDistance/trips MeanTravelDistancePerPerson = totalTravelDistance/pop MeanTravelTime = totalTravelTime/trips MeanTravelTimePerPerson = totalTravelTime/pop return totalTravelDistance, totalTravelTime, MeanTripDistance, MeanTravelDistancePerPerson, MeanTravelTime, MeanTravelTimePerPerson def calculate_relative_mode_share(trip_matrices): """ Calculate the relative mode share for a dictionary of trip matrices. Args: trip_matrices (dict of np.ndarray or pd.DataFrame): Dictionary of trip matrices. Returns: dict: Relative mode distribution for each key in trip_matrices. """ # Compute the total trips for each mode total_trips_per_mode = {key: matrix.sum().sum() for key, matrix in trip_matrices.items()} # Compute the total trips across all modes total_trips_all_modes = sum(total_trips_per_mode.values()) # Calculate the relative mode distribution rel_mode_distribution = {key: trips_per_mode / total_trips_all_modes for key, trips_per_mode in total_trips_per_mode.items()} return rel_mode_distribution def extract_distance_matrices(stream, distance_matrices_of_interest): """ Extract distance matrices from the stream and convert them to pandas DataFrames. Args: stream (dict): Stream data containing distance matrices. distance_matrices_of_interest (list of str): List of keys for the distance matrices of interest. Returns: dict: A dictionary of pandas DataFrames, where each key is a distance matrix kind. """ distance_matrices = {} for distMK in distance_matrices_of_interest: for distM in stream["@Data"]['@{0}']: #print( distM.__dict__.keys()) try: distMdict = distM.__dict__[distMK] distance_matrix_dict = json.loads(distMdict) origin_ids = distance_matrix_dict["origin_uuid"] destination_ids = distance_matrix_dict["destination_uuid"] distance_matrix = distance_matrix_dict["matrix"] # Convert the distance matrix to a DataFrame df_distances = pd.DataFrame(distance_matrix, index=origin_ids, columns=destination_ids) distance_matrices[distMK] = df_distances except Exception as e: pass return distance_matrices #================================================================================================== def computeTrips( df_distributionMatrix, df_total_trips_allNodes, df_distMatrix_speckle, df_alphas, df_attraction_proNode_sum_total, df_distances_aligned, TARGET_TRIP_RATE, SCALING_FACTOR, total_population, tot_res, tot_vis, distance_matrices, metric_matrices, redistributeTrips, DISTANCE_BRACKETS, alpha_low, alpha_med, alpha_high, alpha_mode, alpha_uniform, NEW_F_VALUES, CLIENT, TARGET_STREAM, TARGET_BRANCH, sourceInfo="", ): NEW_ALPHAS = reconstruct_dataframe(alpha_low, alpha_med, alpha_high, df_alphas) NEW_MODE_ALPHA = alpha_mode # ==== #step 1 distributionMatrix_step1M = step_1(df_distributionMatrix, df_total_trips_allNodes) #step 2 df_step_2M = step_2_vectorized(df_distMatrix_speckle, NEW_ALPHAS) #step 3 distAndAreaBasedAttr_step3M, distAndAreaBasedAttr_step3_sumM = step_3_numba(df_attraction_proNode_sum_total, df_step_2M) #step 4 distAndAreaBasedAttr_step4M, distAndAreaBasedAttr_step4_sumM = step_4_numba(distAndAreaBasedAttr_step3M, distAndAreaBasedAttr_step3_sumM) #step 5 df_trips_proNode_proActivity_totalM = step_5_numba(distributionMatrix_step1M, distAndAreaBasedAttr_step4M) #step 6 & 7 df_tripMatrixM = step_6_7_vectorized(df_trips_proNode_proActivity_totalM) #step 8 df_tripMatrix_adjustedM, df_tripMatrix_adjusted_scaledM = adjTripRate_adjFactor(df_tripMatrixM, total_population, TARGET_TRIP_RATE, SCALING_FACTOR ) #------ #MAIN 1 compute trip matrice per mode trip_matricesM = computeModalShare(df_tripMatrix_adjusted_scaledM, distance_matrices, NEW_MODE_ALPHA, f_values=NEW_F_VALUES) #MAIN 2 compute modal shares (redistribute trips in case of identical travel time) trip_matrices_redisM = redistributeModalShares(distance_matrices, trip_matricesM, redistributeTrips) #POST 1 compute mode shares rel_mode_distributionM = calculate_relative_mode_share(trip_matrices_redisM) #POST 2 distance brackets dist_sharesM = computeDistanceBrackets(trip_matrices_redisM, metric_matrices, DISTANCE_BRACKETS) #POST 3 compute more stats (totalTravelDistance, totalTravelTime, MeanTripDistance,MeanTravelDistancePerPerson, MeanTripTime, MeanTravelTimePerPerson) = computeTripStats(trip_matrices_redisM, distance_matrices, metric_matrices, total_population) uniform_tripmatrix = getUniformMatrix(df_distances_aligned, df_tripMatrix_adjustedM, alpha_uniform) #add to dataframe # Define your parameter and target values newdata = { # Model Parameter== # Alpha - Routing "alpha_low": alpha_low, "alpha_med": alpha_med, "alpha_high": alpha_high, "alpha_uniform":alpha_uniform, "fvalues":NEW_F_VALUES, "alpha_mode":NEW_MODE_ALPHA, # Model Indicators == # Modal Shares "share_ped_mm_art": rel_mode_distributionM['activity_node+distance_matrix_ped_mm_art_noEntr'], "share_ped_mm": rel_mode_distributionM['activity_node+distance_matrix_ped_mm_noEntr'], "share_ped": rel_mode_distributionM['activity_node+distance_matrix_ped_noEntr'], "share_ped_art": rel_mode_distributionM['activity_node+distance_matrix_ped_art_noEntr'], # Tripshares by Distance Brackets "800": dist_sharesM["800"], "2400": dist_sharesM["2400"], "4800": dist_sharesM["4800"], # Travel Time & Distances "totalTravelDistance":totalTravelDistance, "totalTravelTime":totalTravelTime, "MeanTravelTimePerPerson":MeanTravelTimePerPerson, # Trip Distances "MeanTripDistance":MeanTripDistance, "MeanTripTime":MeanTripTime, "MeanTravelDistancePerPerson":MeanTravelDistancePerPerson, } trip_matrice_adjName = {k.replace("distance", "trip"):v for k, v in trip_matricesM.items()} trip_matrice_adjName["tripMatrix_landuse"] = df_tripMatrix_adjusted_scaledM trip_matrice_adjName["tripMatrix_uniform"] = uniform_tripmatrix extraData = {"population":total_population, "residents":tot_res, "visitors":tot_vis, "parameter":newdata, } commitMsg = "HuggingFace update" try: commitMsg += " using these commits: #+ " for k,v in sourceInfo.items(): commitMsg += f" {k}: {v}" except: pass print(commitMsg) commit_id = send_matrices_and_create_commit( trip_matrice_adjName, CLIENT, TARGET_STREAM, TARGET_BRANCH, commitMsg, rows_per_chunk=300, containerMetadata=extraData ) print ("===============================") return newdata #================================================================================================== # speckle send def send_matrices_and_create_commit(matrices, client, stream_id, branch_name, commit_message, rows_per_chunk, containerMetadata): transport = ServerTransport(client=client, stream_id=stream_id) matrix_ids = {} container_object = Base() # add to container object for k, v in containerMetadata.items(): container_object[k] = v #initialize the keys() for k, df in matrices.items(): container_object[k] = Base() for k, df in matrices.items(): matrix_object = Base(metaData="Some metadata") # Prepare rows and indices as a single list for each, to be automatically chunked rows = [[round(r,4) for r in row.tolist()] for _, row in df.iterrows()] indices = [str(index) for index in df.index] container_object[k][f"@({rows_per_chunk})"] = rows container_object[k]["indices"] = indices container_objectid = operations.send(base=container_object, transports=[transport]) #container_id = operations.send(base=container_object, transports=[transport]) commit_id = client.commit.create(stream_id=stream_id, object_id=container_objectid, branch_name=branch_name, message=commit_message) return commit_id