def main(input_path="ubicaciones.csv", balance_deviations=[0.1, 0.15, 0.2, 0.3]): """ Use different balance deviations to test complex lp function to minimize. Args: input_path : csv path with information regarding agencies, frequency, volume and coordinates balance_deviations = list with percentage of deviations, for calaculation refer to stops_gap and items_gap Returns: None """ df = pd.read_csv(input_path) df.loc[df[df["Vol_Entrega"] == 0].index, "Vol_Entrega"] = 1 zones = ["D1", "D2", "D3", "D4", "D5", "D6"] agencies = list("A" + df["Id_Cliente"].astype(str)) vol_delivery = list(df["Vol_Entrega"]) vol_stores = list(df["Vol_Entrega"] * df["Frecuencia"]) frequency = list(df["Frecuencia"]) stores_volume = dict(zip(agencies, vol_stores)) stores_frequency = dict(zip(agencies, frequency)) vol_delivery = dict(zip(agencies, vol_delivery)) scaler = MinMaxScaler() fitted_scaler = scaler.fit(df[["lat", "lon"]]) scaled_coordinates = fitted_scaler.transform(df[["lat", "lon"]]) kmeans = KMeansConstrained(n_clusters=6, size_min=604, size_max=605, random_state=12, n_init=100, max_iter=200, n_jobs=-1) kmeans_values = kmeans.fit(scaled_coordinates) df["kmeans"] = list(kmeans.predict(scaled_coordinates)) vectorized_lat_lon = df[["lat", "lon"]].to_numpy() cluster_centers = fitted_scaler.inverse_transform(kmeans.cluster_centers_) distance_matrix = cdist(cluster_centers, vectorized_lat_lon, metric="cityblock") routes = [(z, a) for z in zones for a in agencies] distances = pulp.makeDict([zones, agencies], distance_matrix, 0) flow = pulp.LpVariable.dicts("Distribution", (zones, agencies), 0, None) using = pulp.LpVariable.dicts("BelongstoZone", (zones, agencies), 0, 1, pulp.LpInteger) for percentage in balance_deviations: prob = pulp.LpProblem("BrewingDataCup2020_" + str(percentage), pulp.LpMinimize) prob += pulp.lpSum([ distances[z][a] * flow[z][a] for (z, a) in routes ]) + pulp.lpSum([distances[z][a] * using[z][a] for (z, a) in routes]), "totalCosts" stops_upper, stops_lower = stops_gap(percentage) distr_upper, distr_lower = items_gap(percentage) for z in zones: prob += pulp.lpSum([using[z][a] for a in agencies ]) <= stops_upper, "SumStopsInZoneUpper %s" % z prob += pulp.lpSum([using[z][a] for a in agencies ]) >= stops_lower, "SumStopsInZoneLower %s" % z prob += pulp.lpSum([flow[z][a] for a in agencies ]) <= distr_upper, "SumDistrInZoneUpper %s" % z prob += pulp.lpSum([flow[z][a] for a in agencies ]) >= distr_lower, "SumDistrInZoneLower %s" % z for z in zones: for a in agencies: prob += flow[z][a] - (100000 * using[z][a]) <= 0 prob += flow[z][a] <= vol_delivery[a] for a in agencies: prob += pulp.lpSum([flow[z][a] for z in zones ]) >= stores_volume[a], "Distribution %s" % a prob += pulp.lpSum([ using[z][a] for z in zones ]) == stores_frequency[a], "FrequencyDistribution %s" % a prob.writeLP("lp_files/milp_brewing_" + str(percentage) + ".lp") solver = pulp.CPLEX_CMD(path=path_to_cplex) prob.solve(solver) print("Estado: ", pulp.LpStatus[prob.status]) print("Total Cost: ", pulp.value(prob.objective)) final_df = pd.DataFrame(columns=["D1", "D2", "D3", "D4", "D5", "D6"], index=(range(1, 3626))) final_distr = dict() for v in prob.variables(): if (v.name).find("BelongstoZone_") == 0: if v.varValue > 0: dist = v.name[14:] zone = dist[:2] id_cliente = int(dist[4:]) final_df.loc[id_cliente, zone] = 1 final_df.fillna(0, inplace=True) final_df = final_df.astype(int).reset_index().rename( columns={"index": "Id_Cliente"}) final_df.to_csv("lp_solutions/cplex_opt_" + str(percentage) + "_" + str(pulp.value(prob.objective)) + ".csv", header=True, index=False)
def test_float_precision(): km = KMeansConstrained(n_init=1, random_state=30) inertia = {} X_new = {} centers = {} for dtype in [np.float64, np.float32]: X_test = X.astype(dtype) km.fit(X_test) # dtype of cluster centers has to be the dtype of the input # data assert_equal(km.cluster_centers_.dtype, dtype) inertia[dtype] = km.inertia_ X_new[dtype] = km.transform(X_test) centers[dtype] = km.cluster_centers_ # ensure the extracted row is a 2d array assert_equal(km.predict(X_test[:1]), km.labels_[0]) if hasattr(km, 'partial_fit'): km.partial_fit(X_test[0:3]) # dtype of cluster centers has to stay the same after # partial_fit assert_equal(km.cluster_centers_.dtype, dtype) # compare arrays with low precision since the difference between # 32 and 64 bit sometimes makes a difference up to the 4th decimal # place assert_array_almost_equal(inertia[np.float32], inertia[np.float64], decimal=4) assert_array_almost_equal(X_new[np.float32], X_new[np.float64], decimal=4) assert_array_almost_equal(centers[np.float32], centers[np.float64], decimal=4)
def test_predict(): km = KMeansConstrained(n_clusters=n_clusters, random_state=42) km.fit(X) # sanity check: predict centroid labels pred = km.predict(km.cluster_centers_) assert_array_equal(pred, np.arange(n_clusters)) # sanity check: re-predict labeling for training set samples pred = km.predict(X) assert_array_equal(pred, km.labels_) # re-predict labels for training set using fit_predict pred = km.fit_predict(X) assert_array_equal(pred, km.labels_)
from k_means_constrained import KMeansConstrained df=pd.read_csv("https://raw.githubusercontent.com/JavierLilly/Proyecto_Eco/main/BDC_DATA.csv") #Estandarizando las coordenadas data= df[['lat','lon']].values.astype('float32',copy=False) scaler = StandardScaler().fit(data) data_scal = scaler.transform(data) df_ = df df_[['lat','lon']]=data_scal #Construyendo el modelo de clustering min - max size coor = df_[['lat','lon']] model = KMeansConstrained(n_clusters=6,size_min=600,size_max=700,random_state=5565280).fit(coor) y = model.predict(coor) # Predicion df_['cluster'] = y #Gráfica Todos con Frecuencia >=1 cdict={0:'red',1:'black',2:'yellow',3:'green',4:'blue',5:'grey'} plt.figure(figsize=(10,10)) sns.set() for g in np.unique(y): plt.scatter(coor['lat'][y==g], coor['lon'][y==g], c = cdict[g], label = g, s = 60) # plt.scatter(df['lat'][df['Frecuencia']==2],df['lon'][df['Frecuencia']==2],c='purple',s=80,alpha = .5) # plt.scatter(df['lat'][df['Frecuencia']==3],df['lon'][df['Frecuencia']==3],c='brown',s=150,) plt.legend() #Reducimos los datos
# -*- coding: utf-8 -*- """ Created on Sat Oct 17 19:13:48 2020 @author: lcota """ from k_means_constrained import KMeansConstrained clf = KMeansConstrained(n_clusters=2, size_min=2, size_max=5, random_state=0) clf.fit(X) clf.cluster_centers_ clf.predict([[0, 0], [4, 4]])