Update algoritm.py

algoritm.py

@@ -1,93 +1,118 @@
-
-import os
-import statsmodels.api as sm
-from sklearn.linear_model import LinearRegression
-import pandas as pd
-import numpy as np
-import os
-import shutil
-
-  
-def algoritmo(json_data):
-    d = json_data
-    df_s = pd.DataFrame({
-        "deviceId": d["deviceId"],
-        "TELEMETRY_SEQ_GROUP_ID": d["TELEMETRY_SEQ_GROUP_ID"],
-        **{k: d[k] for k in d if isinstance(d[k], list)}
-    })
-    cols = [
-        'OAT', 'IAS', 'ALPHAI', 'BETAI', 'TETA', 'PHI', 'HDG', 'P', 'Q',
-        'R', 'NR', 'NP1', 'NG1', 'VZ', 'ZRS',
-        'E1A_TM_TOT', 'E1A_TM_T1', 'NGR1', 'DP1'
-    ]
-    X = df_s[cols]
-    assert list(X.columns) == cols
-
-    X = df_s[cols]
-    X = sm.add_constant(X)
-    y_pred = res_loaded.predict(X)
-    y_pred_100 = y_pred * 100
-    df_s['y_pred_100']=y_pred_100
-    #print(len(y_pred_100))
-    df_s["clogging_LH_Predicted"] = np.take(values, np.searchsorted(bins, df_s["y_pred_100"], side="right"))
-    mean_clogging = df_s["clogging_LH_Predicted"].mean()
-    #creo la cartella se non esiste
-    device_id = d["deviceId"]
-    csv_path = os.path.join(device_id, "clogging_mean_by_group.csv")
-    group_id = d["TELEMETRY_SEQ_GROUP_ID"]
-    if group_id == 0:
-        if os.path.exists(device_id):
-            shutil.rmtree(device_id)
-        os.makedirs(device_id)
-    #dataframe da scrivere
-    row = pd.DataFrame([{
-        "TELEMETRY_SEQ_GROUP_ID": group_id,
-        "mean_clogging_LH_Predicted": mean_clogging
-    }])
-    row.to_csv(
-        csv_path,
-        mode="a",
-        header=not os.path.exists(csv_path),
-        index=False
-    )
-    #regressione
-    dtf=pd.read_csv(csv_path)
-    #print(dtf)
-    l=len(dtf)
-    if l>5:
-        X=dtf[['TELEMETRY_SEQ_GROUP_ID']]
-        y=dtf['mean_clogging_LH_Predicted']
-        model = LinearRegression()
-        model.fit(X, y)
-
-        a = model.coef_[0]
-        b = model.intercept_
-
-        x_at_100 = ((100 - b) / a) if a != 0 else np.nan
-        if a != 0:
-          x_at_100 =x_at_100 - int(group_id)
-          slope= a
-          intercept= b
-          r2= model.score(X, y)
-          n_points= l
-        else:
-          x_at_100 =np.nan
-          slope=np.nan
-          intercept= np.nan
-          r2= np.nan
-          n_points= l
-
-    else:
-      x_at_100 =np.nan
-      slope=np.nan
-      intercept= np.nan
-      r2= np.nan
-      n_points= l
-    
-    pic={}
-    pic['x_at_100']=x_at_100
-    pic['slope']=slope
-    pic['intercept']=intercept
-    pic['r2']=r2
-    pic['n_points']=l
+
+import os
+import statsmodels.api as sm
+from sklearn.linear_model import LinearRegression
+import pandas as pd
+import numpy as np
+import os
+import shutil
+
+res_loaded = sm.load("beta_model_clogging.pkl")
+#Prendiamo tutti i limiti superiori, tranne l’ultimo che va a +∞.
+bins = [
+    6.827491567, 7.207134976, 7.507514601, 7.85202832, 8.135308095,
+    8.689867231, 9.286871296, 9.617912395, 10.09183855, 10.73296965,
+    11.10692936, 11.69554997, 12.28992645, 12.73162955, 13.58962501,
+    14.05211378, 14.82733437, 15.28453334, 15.97759006, 16.82037951,
+    17.64491222, 18.42719281, 19.35467075, 20.11238201, 21.11239952,
+    21.8749792, 23.13365381, 24.06277248, 24.69985079, 26.06401012,
+    26.99098886, 28.06155461, 29.31860042, 30.62386734, 31.4413642,
+    32.92566161, 33.97571072, 35.31251088, 36.15053542, 37.57770049,
+    39.13652821, 40.29113449, 41.76522545, 42.82082043, 44.00697262,
+    45.92880868, 46.82432879, 48.63647026, 49.74936819, 51.22777855,
+    52.49036854, 53.51554916, 55.2477514, 56.28299801, 57.51382218,
+    59.16053699, 60.15795894, 61.6664039, 62.90702101, 63.87127815,
+    64.98319564, 66.22208356, 66.61521227, 68.15750081, 69.44821662,
+    70.66864959, 71.92308181, 72.69082311, 73.92476305, 74.87935242,
+    75.77031936, 76.94368679, 78.04134047, 79.02976362, 79.56238123,
+    80.1200946, 81.16954675, 82.1616298, 82.70600553, 83.32428364,
+    84.40737139, 85.17567459, 85.61323671, 86.27587966, 87.00108867,
+    87.48195456, 87.87995156, 88.53346862, 88.95068965, 89.6043336,
+    90.05141954, 90.51239833, 90.99974806, 91.33113872, 91.79738573,
+    92.19314603, 92.59169911, 92.84177084, 93.23587096, np.inf
+]
+values = [0] + list(range(2, 101))
+  
+def algoritmo(json_data):
+    d = json_data
+    df_s = pd.DataFrame({
+        "deviceId": d["deviceId"],
+        "TELEMETRY_SEQ_GROUP_ID": d["TELEMETRY_SEQ_GROUP_ID"],
+        **{k: d[k] for k in d if isinstance(d[k], list)}
+    })
+    cols = [
+        'OAT', 'IAS', 'ALPHAI', 'BETAI', 'TETA', 'PHI', 'HDG', 'P', 'Q',
+        'R', 'NR', 'NP1', 'NG1', 'VZ', 'ZRS',
+        'E1A_TM_TOT', 'E1A_TM_T1', 'NGR1', 'DP1'
+    ]
+    X = df_s[cols]
+    assert list(X.columns) == cols
+
+    X = df_s[cols]
+    X = sm.add_constant(X)
+    y_pred = res_loaded.predict(X)
+    y_pred_100 = y_pred * 100
+    df_s['y_pred_100']=y_pred_100
+    #print(len(y_pred_100))
+    df_s["clogging_LH_Predicted"] = np.take(values, np.searchsorted(bins, df_s["y_pred_100"], side="right"))
+    mean_clogging = df_s["clogging_LH_Predicted"].mean()
+    #creo la cartella se non esiste
+    device_id = d["deviceId"]
+    csv_path = os.path.join(device_id, "clogging_mean_by_group.csv")
+    group_id = d["TELEMETRY_SEQ_GROUP_ID"]
+    if group_id == 0:
+        if os.path.exists(device_id):
+            shutil.rmtree(device_id)
+        os.makedirs(device_id)
+    #dataframe da scrivere
+    row = pd.DataFrame([{
+        "TELEMETRY_SEQ_GROUP_ID": group_id,
+        "mean_clogging_LH_Predicted": mean_clogging
+    }])
+    row.to_csv(
+        csv_path,
+        mode="a",
+        header=not os.path.exists(csv_path),
+        index=False
+    )
+    #regressione
+    dtf=pd.read_csv(csv_path)
+    #print(dtf)
+    l=len(dtf)
+    if l>5:
+        X=dtf[['TELEMETRY_SEQ_GROUP_ID']]
+        y=dtf['mean_clogging_LH_Predicted']
+        model = LinearRegression()
+        model.fit(X, y)
+
+        a = model.coef_[0]
+        b = model.intercept_
+
+        x_at_100 = ((100 - b) / a) if a != 0 else np.nan
+        if a != 0:
+          x_at_100 =x_at_100 - int(group_id)
+          slope= a
+          intercept= b
+          r2= model.score(X, y)
+          n_points= l
+        else:
+          x_at_100 =np.nan
+          slope=np.nan
+          intercept= np.nan
+          r2= np.nan
+          n_points= l
+
+    else:
+      x_at_100 =np.nan
+      slope=np.nan
+      intercept= np.nan
+      r2= np.nan
+      n_points= l
+    
+    pic={}
+    pic['x_at_100']=x_at_100
+    pic['slope']=slope
+    pic['intercept']=intercept
+    pic['r2']=r2
+    pic['n_points']=l
     return pic