我可以将 Tensorboard 用于我的线性回归或线性分类问题吗？

Question

I am investigating whether I can use tensorboard in my project. In my project I use data (csv file) and I don't know if I can use tensorboard for loss, predicts or if it can only be used for image recognition. I do not really manage it myself. does anyone have an idea if this is possible and how you could do this? i use tensorflow.

Thanks.

Answer 1

Yes. You can use Tensorboard for Regression also. You can monitor 'scalars , distributions , graphs , histogram` etc to monitor how parameters are getting updated over epochs or time.

I have used simple example and shown how to usage in colab.

import tensorflow as tf
import numpy as np
from tensorflow import keras
import os
import datetime
model = tf.keras.Sequential([keras.layers.Dense(units=1, input_shape=[1])])
model.compile(optimizer='sgd', loss='mean_squared_error')
xs = np.array([-1.0,  0.0, 1.0, 2.0, 3.0, 4.0], dtype=float)
ys = np.array([-3.0, -1.0, 1.0, 3.0, 5.0, 7.0], dtype=float) # 2*x-1
logdir = os.path.join("logs", datetime.datetime.now().strftime("%Y%m%d-%H%M%S"))
tensorboard_callback = tf.keras.callbacks.TensorBoard(logdir, histogram_freq=1)
model.fit(xs, ys, epochs=500, verbose=0,callbacks=[tensorboard_callback])
print(model.predict([6.0])) # actual value is 11.0 and predicts 10.99

%load_ext tensorboard
%tensorboard --logdir logs

Full code is here for reference.

Answer 2

Thanks for your answer. it works well!

But now I have an another problem. I want to use a csv file like this: import tensorflow as tf import numpy as np from tensorflow import keras import pandas as pd import sklearn from sklearn import linear_model

import os
from datetime import datetime

data = pd.read_csv("student-mat.csv", sep=";")
data = data[["G1", "G2", "G3", "studytime", "failures", "absences"]]
predict = "G3"

model = tf.keras.Sequential([
keras.layers.Dense(units=5, input_shape=[5])])
model.compile(optimizer='sgd',
          loss='mean_squared_error',
          metrics=['accuracy'])

xs = np.array(data.drop([predict], 1))
ys = np.array(data[predict])

x_train, x_test, y_train, y_test =         sklearn.model_selection.train_test_split(xs, ys, test_size=0.1)

log_dir= ".\\tensorflow_logs\\test\\"+datetime.now().strftime("%Y%m%d-    %H%M%S")
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir,     histogram_freq=1)

model.fit(x_train,
      y_train,
      epochs=500,
      verbose=0,
      validation_data=(x_test, y_test),
      callbacks=[tensorboard_callback])

I get the error messages:

ValueError: A target array with shape (355, 1) was passed for an output of shape (None, 355) while using as loss mean_squared_error . This loss expects targets to have the same shape as the output.

do you now a solution for this problem. i think it has to do with the input_shape but idk?