[M] bug fixed: signature, loading sign, outliers+rural

README.md

@@ -12,6 +12,7 @@ Finding a real estate in a new city is still a challenge. We often arrive in a c
 
 - Python, version >=3.8
 - [MongoDB](https://www.mongodb.com/), version >=4 for importing the database about neighbourhoods.
+- We recommend to run Predihood on Firefox or Chrome (Safari get some issues) 
 
 ### Installation
 
@@ -28,7 +29,7 @@ This command install dependencies, including [mongiris](https://gitlab.liris.cnr
 Next, to install the database, run the MongoDB server and execute this command (from the MongoDB's executables directory if needed):
 
 ```
-./mongorestore --archive=/path/to/dump-iris.bin
+./mongorestore --archive=/path/to/dump-dbinsee.bin
 ```
 
 where `/path/to/` is the path to the dump file of the IRIS collection (provided with the package mongiris in `mongiris/data/dump-iris.bin`). 

predihood/classes/Dataset.py

@@ -41,7 +41,7 @@ class Dataset:
         self.Y_test = None
         if env in ENVIRONMENT_VARIABLES: self.env = env
         else: self.env = "building_type"
-        self.train_size, self.test_size = check_train_test_percentages(train_size, test_size)
+        self.train_size, self.test_size, _ = check_train_test_percentages(train_size, test_size)
         self.outliers = outliers
         self.remove_rural = remove_rural
 
@@ -51,11 +51,15 @@ class Dataset:
         When the type is "unsupervised", split data into X and Y is not relevant (as there is no train/test sets).
         """
         if self.type == "supervised":
-            if self.outliers: self.remove_outliers()
             if self.remove_rural: self.remove_rural_iris()
             self.init_X()
             self.init_Y()
             self.train_test()
+            if self.outliers:
+                self.remove_outliers()  # after train_test() because Isolation Forest needs X_train and X_test
+                self.init_X()
+                self.init_Y()
+                self.train_test()  # need to compute train_test to update X and Y after dropping outliers
 
     def init_X(self):
         """
@@ -100,10 +104,9 @@ class Dataset:
         isolation_forest.fit(self.X_train)
         predictions = isolation_forest.predict(self.X_test)
         for i in range(len(predictions)):
-            if predictions[i] == -1:
-                code = self.data.loc[i, "CODE"]
-                log.debug(code, "has been removed since it is an outlier.")
-                self.data = self.data.drop(i, axis=0)  # delete IRIS that are detected as outliers
+            if predictions[i] == -1:  # this IRIS is an outlier according to isolation forest algorithm
+                # delete IRIS that are detected as outliers, ignore id the iris is not present in the dataset (e.g. because it has been remove as a rural IRIS)
+                self.data = self.data.drop(i, axis=0, errors="ignore")
 
     def remove_rural_iris(self):
         """

predihood/main.py

@@ -17,7 +17,7 @@ from predihood.classes.Data import Data
 from predihood.classifiers_list import AVAILABLE_CLASSIFIERS
 from predihood.config import FILE_MANUAL_ASSESSMENT_EXPERTS
 from predihood.predict import compute_all_accuracies, predict_one_iris
-from predihood.utility_functions import signature, get_classifier, set_classifier, add_assessment_to_file
+from predihood.utility_functions import signature_of_algorithm, get_classifier, set_classifier, add_assessment_to_file
 from sklearn.utils._testing import ignore_warnings
 
 log = logging.getLogger(__name__)
@@ -90,7 +90,11 @@ def get_classifiers():
     Returns:
          A list containing the names of the available classifiers.
     """
-    return {"classifiers": list(AVAILABLE_CLASSIFIERS.keys())}
+    available_classifiers = list(AVAILABLE_CLASSIFIERS.keys())
+    if available_classifiers is not None:
+        return {"classifiers": available_classifiers}
+    else:
+        return {"error": "The list of classifiers can not be retrieved. You might reload the page."}
 
 
 @app.route('/getParameters', methods=["GET"])
@@ -103,10 +107,14 @@ def get_parameters():
     """
     if 'name' in request.args:
         name = request.args['name']
-        parameters = signature(name)
-        return parameters
+        parameters, message = signature_of_algorithm(name)
+        if parameters != {} and message == "OK":
+            return parameters
+        else:
+            if message == "no_documentation": return "There is no documentation for the selected algorithm, therefore parameters can not be retrieved."
+            elif message == "no_parameters": return "There was a problem while retrieving algorithm's parameters. You can check the Python's console."
     else:
-        return None
+        return "There is no parameters to get. Be sure to select an algorithm in the list."
 
 
 @app.route('/run', methods=["GET"])
@@ -134,8 +142,8 @@ def run_algorithm():
     # 3. run experiment on data to get accuracies for each EV and each list of selected indicators
     data = Data(normalization="density", filtering=True, add_assessment=add_manual_assessment)
     data.init_all_in_one()
-    accuracies = compute_all_accuracies(data, clf, train_size, test_size, remove_outliers, remove_rural)
-    return {"results": accuracies, "tops_k": predihood.config.TOPS_K}
+    accuracies, message = compute_all_accuracies(data, clf, train_size, test_size, remove_outliers, remove_rural)
+    return {"results": accuracies, "tops_k": predihood.config.TOPS_K, "message": message}
 
 
 @app.route('/predict_iris', methods=["GET"])
@@ -152,8 +160,8 @@ def predict_iris():
 
     data = Data(normalization="density", filtering=True)
     data.init_all_in_one()
-    predictions = predict_one_iris(iris_code_to_predict, data, clf, 0.8, 0.2, False)  # clf
-    return {"predictions": predictions}
+    predictions, message = predict_one_iris(iris_code_to_predict, data, clf, 0.8, 0.2, False)  # clf
+    return {"predictions": predictions, "message": message}
 
 
 @app.route('/getIrisPolygon', methods=["GET"])

predihood/predict.py

@@ -47,7 +47,7 @@ def compute_all_accuracies(data, clf, train_size, test_size, remove_outliers=Fal
         a dictionary of results for each EV and each list of selected indicators
     """
     log.info("... Computing accuracies ...")
-    train_size, test_size = check_train_test_percentages(train_size, test_size)
+    train_size, test_size, message = check_train_test_percentages(train_size, test_size)
 
     data_not_filtered = Data(normalization="density", filtering=False, add_assessment=data.add_assessment)
     data_not_filtered.init_all_in_one()
@@ -98,7 +98,7 @@ def compute_all_accuracies(data, clf, train_size, test_size, remove_outliers=Fal
         log.info("mean for classifier: %f", mean_classifier)
     results = OrderedDict(results)
     log.info(results)
-    return results
+    return results, message
 
 
 def predict_one_iris(iris_code, data, clf, train_size, test_size, remove_outliers=False):
@@ -116,7 +116,7 @@ def predict_one_iris(iris_code, data, clf, train_size, test_size, remove_outlier
     Returns:
         A dictionary containing predictions for each EV.
     """
-    train_size, test_size = check_train_test_percentages(train_size, test_size)
+    train_size, test_size, message = check_train_test_percentages(train_size, test_size)
     lists = retrieve_lists()
 
     predictions = {}
@@ -139,7 +139,7 @@ def predict_one_iris(iris_code, data, clf, train_size, test_size, remove_outlier
         else:
             predictions[env] = get_most_frequent(predictions_lst)  # get the most frequent value and the number of occurrences
     print(predictions)  # {'building_type': {'most_frequent': 'Towers', 'count_frequent': 7}, 'building_usage': {'most_frequent': 'Housing', 'count_frequent': 4}, ... }
-    return predictions
+    return predictions, message
 
 
 if __name__ == '__main__':

predihood/static/js/algorithms.js

@@ -58,7 +58,8 @@ $("#testPercentage")
 // run the classifier with specified parameters and display results in the results section.
 $("#runBtn").click("on", function () {
     // $("body").css("cursor", "progress");
-    document.body.classList.add('busy-cursor');
+    // $("html,body").css("cursor", "progress");
+    document.body.style.cursor = 'wait';
     $(".wrapperTable input[type='checkbox']:not(:checked)").each(function () {
         $(this).parent().parent().empty(); // remove tables that are not checked in the interface
     });
@@ -97,7 +98,7 @@ $("#runBtn").click("on", function () {
     request_run = $.ajax({
         "type": "GET",
         "url": "/run",
-        "async": false,
+        //"async": false,
         data: {
             "clf": chosen_clf,
             "parameters": JSON.stringify(userParameters)
@@ -108,6 +109,7 @@ $("#runBtn").click("on", function () {
             //   - the results table, highlighted cells are best means for each EV
             //   - the list of parameters associated to the results
             //   - the mean for this classifier (all EV combined)
+            if(result["message"] === "KO") { alert("Your train_size or test_size parameter was set badly. The algorithm has been run with default parameters which are 80% for train_size and 20% for test_size.") }
             let keep = $("<label class='h5'><input type='checkbox' style='margin-right: 1rem;'/>" + chosen_clf + "</label>");
             let table = $("<table id='tableToExport'>")
                 .addClass("table table-hover table-responsive")
@@ -195,17 +197,15 @@ $("#runBtn").click("on", function () {
 
             // append all to HTML
             $("#resultsDiv").append(containing_table);
-            // $("body").css("cursor", "default");
-            document.body.classList.remove('busy-cursor');
+            document.body.style.cursor = 'default';
         },
         error: function (result, textStatus, errorThrown) {
             console.log(errorThrown);
-            alert("something went wrong while training. Please check your parameters<br>" + textStatus);
-            // $("body").css("cursor", "default");
-            document.body.classList.remove('busy-cursor');
+            alert("Something went wrong while training. Please check your parameters." + textStatus);
+            document.body.style.cursor = 'default';
+
         }
     });
-
     return false; // do not reload
 });
 
@@ -366,15 +366,19 @@ function getParameters(name) {
         },
         contentType: 'application/json;charset=UTF-8',
         success: function (result) {
-            current_parameters = result; // store current parameters (for the chosen classifier) in a global variable
-            let parameters = result;
-            console.log(parameters)
-            if (Object.keys(parameters).length > 0) {
-                // at least one key
-                addParameters(parameters);
+            if(result instanceof String) {
+                alert(result)
             } else {
-                // nothing to append, empty the div (no selection)
-                $("#tuningSection").css("visibility", "hidden");
+                current_parameters = result; // store current parameters (for the chosen classifier) in a global variable
+                let parameters = result;
+                console.log(parameters)
+                if (Object.keys(parameters).length > 0) {
+                    // at least one key
+                    addParameters(parameters);
+                } else {
+                    // nothing to append, empty the div (no selection)
+                    $("#tuningSection").css("visibility", "hidden");
+                }
             }
         },
         error: function (result, textStatus, errorThrown) {

predihood/static/js/carto.js

@@ -78,32 +78,37 @@ function displayPopup(e) {
 
     // common part of the popup
     let divInformation = $("<div>");
+    let download;
+
     if(preferred_language_carto === "french") {
+        download = $("<i class='fas fa-download' style='margin-left: 1rem;' title='Exporter cette table comme un fichier Excel.'></i>");
         divInformation
-            .append("CODE IRIS : " + layer.feature.properties.CODE_IRIS).append($("<br>"))
+            .append("CODE IRIS : " + code_iris).append($("<br>"))
             .append("IRIS : " + layer.feature.properties.NOM_IRIS).append($("<br>"))
             .append("COMMUNE : " + layer.feature.properties.NOM_COM).append($("<br>"))
         let moreInfosLink = $("<a>");
         moreInfosLink
-            .prop("href", "details-iris.html?code_iris="+layer.feature.properties.CODE_IRIS)
+            .prop("href", "details-iris.html?code_iris="+code_iris)
             .prop("target", "_blank")
             .text("Plus de détails")
             .append($("<br>"));
-        divInformation.append(moreInfosLink);
+        divInformation.append(moreInfosLink).append(download);
     } else {
+        download = $("<i class='fas fa-download' style='margin-left: 1rem;' title='Export this table as an Excel file.'></i>");
         divInformation
-            .append("IRIS CODE: " + layer.feature.properties.CODE_IRIS).append($("<br>"))
+            .append("IRIS CODE: " + code_iris).append($("<br>"))
             .append("IRIS: " + layer.feature.properties.NOM_IRIS).append($("<br>"))
-            .append("TOWNSHIP: " + layer.feature.properties.NOM_COM).append($("<br>"))
+            .append("TOWNSHIP: " + layer.feature.properties.NOM_COM).append($("<br>"));
         let moreInfosLink = $("<a>");
         moreInfosLink
-            .prop("href", "details-iris.html?code_iris="+layer.feature.properties.CODE_IRIS)
+            .prop("href", "details-iris.html?code_iris="+code_iris)
             .prop("target", "_blank")
             .text("More details")
             .append($("<br>"));
-        divInformation.append(moreInfosLink);
+        divInformation.append(moreInfosLink).append(download);
     }
 
+
     if (selected_algorithm !== "undefined" && selected_algorithm !== undefined) {
         predictions = predict(code_iris, selected_algorithm)
         console.log(predictions)
@@ -119,11 +124,36 @@ function displayPopup(e) {
     }
     previously_selected_algorithm = selected_algorithm;
 
-    let divPredictions = $("<div>").prop("id", "divPredictions")
-    if(predictions !== undefined) {
-        for(let key in predictions) { divPredictions.append(capitalizeFirstLetter(key.split("_").join(" "))+': ' + predictions[key]["most_frequent"] + " (" + predictions[key]["count_frequent"] + "/7)").append($('<br>')); }
+    let divPredictions = $("<div>").prop("id", "divPredictions");
+    let tablePredictions = $("<table>").prop("id", "tablePredictions");
+    tablePredictions.append("<th><td>Environment value</td><td>Prediction</td><td>Frequency</td></th>")
+    let row, colEnv, colValue, colFrequency;
+    if (predictions !== undefined) {
+        for (let key in predictions) {
+            divPredictions.append(
+                capitalizeFirstLetter(key.split("_").join(" ")) + ': '
+                + predictions[key]["most_frequent"]
+                + " (" + predictions[key]["count_frequent"] + "/7)"
+            ).append($('<br>'));
+            row = $("<tr>");
+            colEnv = $("<td>").html(capitalizeFirstLetter(key.split("_").join(" ")));
+            colValue = $("<td>").html(predictions[key]["most_frequent"]);
+            colFrequency = $("<td>").html(predictions[key]["count_frequent"] + "/7");
+            row.append(colEnv, colValue, colFrequency)
+            tablePredictions.append(row);
+        }
+        console.log(tablePredictions);
+        download.on("click", function (e) {
+            e.preventDefault();
+            tablePredictions.table2excel({
+                type: 'xls',
+                filename: 'predictions'+code_iris+'.xls',
+                preserveColors: true
+            });
+        });
     }
 
+
     let messageTooltip = divInformation[0].outerHTML + selectAlgorithm[0].outerHTML + divPredictions[0].outerHTML;
     console.log(messageTooltip)
     layer.bindPopup(messageTooltip)

predihood/static/js/prediction.js

@@ -17,13 +17,15 @@ function predict(iris_code, algorithm_name) {
 		"async": false,
 		contentType: 'application/json;charset=UTF-8',
 		success: function(result) {
-			console.log(result)
-			console.log(result['predictions'])
+			console.log(result);
+			console.log(result['predictions']);
 			predictions = result['predictions'];
+			if(result['message'] === "KO") { alert("Your train_size or test_size parameter was set badly. The algorithm has been run with default parameters which are 80% for train_size and 20% for test_size."); }
 			$("body").css("cursor", "default");
 		},
 		error: function(result, textStatus, errorThrown) {
             console.log(errorThrown);
+            alert("There is an issue to predict environment of this IRIS. You can check the console to have more details. " + result + textStatus + errorThrown);
             $("body").css("cursor", "default");
 		}
 	});

predihood/tests.py

@@ -10,7 +10,7 @@ import unittest
 from predihood.config import FOLDER_DATASETS, ENVIRONMENT_VALUES
 from predihood.utility_functions import check_train_test_percentages, intersection, union, similarity, \
     get_most_frequent, address_to_code, address_to_city, indicator_full_to_short_label, \
-    indicator_short_to_full_label, get_classifier, set_classifier, signature, add_assessment_to_file
+    indicator_short_to_full_label, get_classifier, set_classifier, signature_of_algorithm, add_assessment_to_file
 from sklearn.neighbors import KNeighborsClassifier
 
 
@@ -69,7 +69,7 @@ class TestCase(unittest.TestCase):
     def test_signature(self):
         # test if the signature function is able to pick correct parameters
         classifier_name = "KNeighbors Classifier"
-        classifier_signature = signature(classifier_name)
+        classifier_signature = signature_of_algorithm(classifier_name)
         assert "n_neighbors" in classifier_signature
         assert classifier_signature["n_neighbors"]["types"] == ["int"]
         assert classifier_signature["n_neighbors"]["default"] == "5"

predihood/utility_functions.py

@@ -2,6 +2,8 @@ import ast
 import inspect
 import json
 import logging
+import sys
+
 import numpy as np
 import pandas as pd
 import re
@@ -9,6 +11,7 @@ import requests
 import stringdist
 
 from area import area
+from inspect import signature
 from predihood import model
 from predihood.classifiers_list import AVAILABLE_CLASSIFIERS
 from predihood.config import TRAIN_SIZE, TEST_SIZE, ENVIRONMENT_VARIABLES, TRANSLATION, OLD_PREFIX, NEW_PREFIX, FILE_MANUAL_ASSESSMENT
@@ -203,7 +206,7 @@ def set_classifier(classifier, parameters):
     return classifier
 
 
-def signature(chosen_algorithm):
+def signature_of_algorithm(chosen_algorithm):
     """
     Get the signature of an algorithm, i.e. its parameters, the default values and the type of each parameter. The documentation of the algorithm must be in NumPy style.
 
@@ -216,11 +219,26 @@ def signature(chosen_algorithm):
         - the default value
         - a description of the parameter (e.g. "The train_size parameter aims at tuning the size of the sample during the learning step.")
     """
-    # special case for no selection
-    if chosen_algorithm == "Algorithm": return json.dumps({})
+    if chosen_algorithm == "Algorithm": return json.dumps({}), "no_selection"  # special case for no selection
+
+    # getting parameters and default values
+    parameters = {}
+    model = get_classifier(chosen_algorithm)
+    if sys.version_info.major < 3 or (sys.version_info.major == 3 and sys.version_info.minor < 8):
+        params = inspect.getfullargspec(model.__init__).args[1:]  # get parameter' names -- [1:] to remove self parameter
+        defaults = inspect.getfullargspec(model.__init__).defaults  # get default values
+    else:
+        # getfullargspec is deprecated since Python 3.8, need to use inspect?signature instead
+        sig = inspect.signature(model.__init__)
+        params, defaults = [], []
+        for param in sig.parameters.values():
+            params.append(param.name)
+            defaults.append(param.default)
+    if params is None or defaults is None: return json.dumps({}), "no_parameters"
+    assert len(params) == len(defaults)
+
+    # getting the definition of parameters
     try:
-        # model = eval(_chosen_algorithm) # never use eval on untrusted strings
-        model = get_classifier(chosen_algorithm)
         doc = model.__doc__
         param_section = "Parameters"
         dashes = "-" * len(param_section)  # -------
@@ -229,11 +247,7 @@ def signature(chosen_algorithm):
         # sub_doc is the param section of the docs (i.e. without attributes and some text)
         sub_doc = doc[doc.find(param_section) + len(param_section) + number_spaces + len(dashes) + len("\n"):doc.find(attribute_section)]
     except:
-        raise Exception("This algorithm does not have a documentation...")
-    params = inspect.getfullargspec(model.__init__).args[1:]  # get parameter' names -- [1:] to remove self parameter
-    defaults = inspect.getfullargspec(model.__init__).defaults  # get default values
-    assert len(params) == len(defaults)
-    parameters = {}
+        return json.dumps({}), "no_documentation"
     for i in range(len(params)):
         param_name = str(params[i]) + " : "
         index_param = sub_doc.find(param_name)
@@ -284,7 +298,7 @@ def signature(chosen_algorithm):
         if len(types) == 0: types.append(type_of_default)  # fill missing types
         types[:] = [x for x in types if "None" not in x and "NoneType" not in x]  # remove None type
         parameters[param_name[:-3]] = {"types": types, "default": default, "description": first_sentence}  # -3 to remove " : "
-    return parameters
+    return parameters, "OK"
 
 
 def check_train_test_percentages(train_size, test_size):
@@ -299,17 +313,21 @@ def check_train_test_percentages(train_size, test_size):
         the train and test sizes
     """
     if 0 < train_size < 1 and 0 < test_size < 1 and train_size + test_size == 1:
-        return train_size, test_size  # default case
+        message = "OK"  # default case - everything is correct
     elif 0 < train_size < 1 and 0 < test_size < 1:
+        message = "KO"  # train_size or test_size is not correct so we set default parameters
         train_size = TRAIN_SIZE  # 0.8
         test_size = TEST_SIZE  # 0.2
 
     if 1 <= train_size < 100 and 1 <= test_size < 100 and train_size + test_size == 100:
-        return train_size / 100, test_size / 100  # default case
+        message = "OK"  # default case - everything is correct
+        train_size = train_size/100
+        test_size = test_size/100
     elif 1 <= train_size < 100 and 1 <= test_size < 100:
+        message = "KO"  # train_size or test_size is not correct so we set default parameters
         train_size = TRAIN_SIZE  # 0.8
         test_size = TEST_SIZE  # 0.2
-    return train_size, test_size
+    return train_size, test_size, message
 
 
 # 2. list functions