diff options
Diffstat (limited to 'backend')
| -rw-r--r-- | backend/api/api/Models/Dataset.cs | 1 | ||||
| -rw-r--r-- | backend/api/api/Models/Model.cs | 3 | ||||
| -rw-r--r-- | backend/api/api/Services/FillAnEmptyDb.cs | 6 | ||||
| -rw-r--r-- | backend/microservice/api/controller.py | 7 | ||||
| -rw-r--r-- | backend/microservice/api/newmlservice.py | 17 |
5 files changed, 23 insertions, 11 deletions
diff --git a/backend/api/api/Models/Dataset.cs b/backend/api/api/Models/Dataset.cs index 7acd4382..beb66de9 100644 --- a/backend/api/api/Models/Dataset.cs +++ b/backend/api/api/Models/Dataset.cs @@ -28,6 +28,7 @@ namespace api.Models public int nullRows { get; set; } public bool isPreProcess { get; set; } + public float[][] cMatrix { get; set; } } } diff --git a/backend/api/api/Models/Model.cs b/backend/api/api/Models/Model.cs index f95be715..a2740ca9 100644 --- a/backend/api/api/Models/Model.cs +++ b/backend/api/api/Models/Model.cs @@ -27,7 +27,8 @@ namespace api.Models public string lossFunction { get; set; } //public int inputNeurons { get; set; } public int hiddenLayers { get; set; } - public int batchSize { get; set; } + public string batchSize { get; set; } + public string learningRate { get; set; } // na izlazu je moguce da bude vise neurona (klasifikacioni problem sa vise od 2 klase) public int outputNeurons { get; set; } public Layer[] layers { get; set; } diff --git a/backend/api/api/Services/FillAnEmptyDb.cs b/backend/api/api/Services/FillAnEmptyDb.cs index 6148efb6..d8c53bab 100644 --- a/backend/api/api/Services/FillAnEmptyDb.cs +++ b/backend/api/api/Services/FillAnEmptyDb.cs @@ -99,7 +99,7 @@ namespace api.Services model.optimizer = "Adam"; model.lossFunction = "mean_squared_error"; model.hiddenLayers = 5; - model.batchSize = 8; + model.batchSize = "8"; model.outputNeurons = 0; model.outputLayerActivationFunction = "sigmoid"; model.metrics = new string[] { }; @@ -212,7 +212,7 @@ namespace api.Services model.optimizer = "Adam"; model.lossFunction = "mean_absolute_error"; model.hiddenLayers = 4; - model.batchSize = 5; + model.batchSize = "8"; model.outputNeurons = 0; model.outputLayerActivationFunction = "relu"; model.metrics = new string[] { }; @@ -320,7 +320,7 @@ namespace api.Services model.optimizer = "Adam"; model.lossFunction = "sparse_categorical_crossentropy"; model.hiddenLayers = 3; - model.batchSize = 4; + model.batchSize = "64"; model.outputNeurons = 0; model.outputLayerActivationFunction = "softmax"; model.metrics = new string[] { }; diff --git a/backend/microservice/api/controller.py b/backend/microservice/api/controller.py index 41035cc7..988ad987 100644 --- a/backend/microservice/api/controller.py +++ b/backend/microservice/api/controller.py @@ -118,7 +118,6 @@ def returnColumnsInfo(): ''' preprocess = newmlservice.returnColumnsInfo(data) #samo 10 jedinstvenih posto ih ima previse, bilo bi dobro da promenimo ovo da to budu 10 najzastupljenijih vrednosti - for col in preprocess["columnInfo"]: col["uniqueValues"] = col["uniqueValues"][0:6] col["uniqueValuesCount"] = col["uniqueValuesCount"][0:6] @@ -128,11 +127,9 @@ def returnColumnsInfo(): dataset["nullRows"] = preprocess["allNullRows"] dataset["colCount"] = preprocess["colCount"] dataset["rowCount"] = preprocess["rowCount"] + dataset["cMatrix"]=preprocess['cMatrix'] dataset["isPreProcess"] = True - #print(dataset) - - - + return jsonify(dataset) print("App loaded.") diff --git a/backend/microservice/api/newmlservice.py b/backend/microservice/api/newmlservice.py index 9e26b03a..f5e5abcc 100644 --- a/backend/microservice/api/newmlservice.py +++ b/backend/microservice/api/newmlservice.py @@ -27,15 +27,28 @@ import matplotlib.pyplot as plt #from ann_visualizer.visualize import ann_viz; def returnColumnsInfo(dataset): dict=[] + datafront=dataset.copy() + dataMatrix=dataset.copy() + + svekolone=datafront.columns kategorijskekolone=datafront.select_dtypes(include=['object']).columns + allNullCols=0 rowCount=datafront.shape[0]#ukupan broj redova colCount=len(datafront.columns)#ukupan broj kolona for kolona in svekolone: if(kolona in kategorijskekolone): + encoder=LabelEncoder() + dataMatrix[kolona]=encoder.fit_transform(dataMatrix[kolona]) + + #print(dataMatrix.dtypes) + cMatrix=dataMatrix.corr() + + for kolona in svekolone: + if(kolona in kategorijskekolone): unique=datafront[kolona].value_counts() uniquevalues=[] uniquevaluescount=[] @@ -86,7 +99,7 @@ def returnColumnsInfo(dataset): #pretvaranje u kategorijsku datafront = datafront.astype({kolona: str}) - print(datafront.dtypes) + #print(datafront.dtypes) unique=datafront[kolona].value_counts() uniquevaluesn=[] uniquevaluescountn=[] @@ -117,7 +130,7 @@ def returnColumnsInfo(dataset): #print(NullRows) #print(len(NullRows)) allNullRows=len(NullRows) - return {'columnInfo':dict,'allNullColl':int(allNullCols),'allNullRows':int(allNullRows),'rowCount':int(rowCount),'colCount':int(colCount)} + return {'columnInfo':dict,'allNullColl':int(allNullCols),'allNullRows':int(allNullRows),'rowCount':int(rowCount),'colCount':int(colCount),'cMatrix':str(np.matrix(cMatrix))} @dataclass class TrainingResultClassification: |