I've built an LSTM In Keras with the goal of predicting future values of a time-series from a high-dimensional, time-index input.
However, there's a unique requirement: for certain time points in the future, we know with certainty what some values of the input series will be. For example:
model = SomeLSTM()
trained_model = model.train(train_data)
known_data = [(24, {feature: 2, val: 7.0}), (25, {feature: 2, val: 8.0})]
predictions = trained_model(look_ahead=48, known_data=known_data)
Which would train the model up to time t (the end of training), and predict forward 48 time periods from time t, but substituting known_data values for feature 2 at times 24 and 25.
How exactly can I explicitly inject this into the LSTM at some time?
For reference, here's the model:
model = Sequential()
model.add(LSTM(hidden, input_shape=(look_back, num_features)))
model.add(Dropout(dropout))
model.add(Dense(look_ahead))
model.add(Activation('linear'))
This may be a result of my un-intuitive grasp of LSTMs, and I'd appreciate any clarification. I've dived into the Keras source code, and my first guess is to inject it right into the LSTM state variable, but I'm unsure how to do that at time t (or even if that is correct.)
I think a clean way of doing this is to introduce 2*look_ahead new features, where for each 0 <= i < look_ahead 2*i-th feature is an indicator whether the value of the i-th time step is known and (2*i+1)-th is the value itself (0 if not known). Accordingly, you can generate training data with these features to make your model take into account these known values.
I am not exactly sure what you are trying to do, but maybe create your own layer to go at the end that sets the data to the known values, similar to how dropout sets random values to zero. As a side note, I have had better results with pooling than dropout, so maybe try switching that out and training it. Here is a good guide on how to do it. https://www.tutorialspoint.com/keras/keras_customized_layer.htm
Related
So I have an output vector of dim=7 and 4 possible classes for each position, so my question is, is it possible to feed the keras model a vector of one hot vectors, where each position of the vector is a one hot vector? something like this [[1000],[1000],[0100],[0010],[0001],[0001],[0010]].
If this is not possible are there any alternatives?
If you want your output of your model to be like that when your model = keras.models.Model(...), the answer is not possible because the output that you provide (which is like a step respond "[1000] => [0000]") will have a gradient of infinity at step and 0 at other point.
What people do is to create a model that give distribution over different action and select the highest probability as predicted value and using cross entropy loss to optimize the model. For example, from your output [1,0,0,0] you will have something like [0.9,0.01,0.01,0.08] instead. Then you can pick first instance as predicted value. This will make sure that your model have gradient at all point.
So if you really want your model to have dim = 7 and 4 different value, you can create output size of 28 = 7*4 with sigmoid activation, then pick first 4 as your dimension 1 (maybe using something like np.argmax(output[0:4])) and so on.
I am developing a model using linear regression to predict the age. I know that the age is from 0 to 100 and it is a possible value. I used conv 1 x 1 in the last layer to predict the real value. Do I need to add a ReLU function after the output of convolution 1x1 to guarantee the predicted value is a positive value? Currently, I did not add ReLU and some predicted value becomes negative value like -0.02 -0.4…
There's no compelling reason to use an activation function for the output layer; typically you just want to use a reasonable/suitable loss function directly with the penultimate layer's output. Specifically, a RELU doesn't solve your problem (or at most only solves 'half' of it) since it can still predict above 100. In this case -predicting a continuous outcome- there's a few standard loss functions like squared error or L1-norm.
If you really want to use an activation function for this final layer and are concerned about always predicting within a bounded interval, you could always try scaling up the sigmoid function (to between 0 and 100). However, there's nothing special about sigmoid here - any bounded function, ex. any CDF of a signed, continuous random variable, could be similarly used. Though for optimization, something easily differentiable is important.
Why not start with something simple like squared-error loss? It's always possible to just 'clamp' out-of-range predictions to within [0-100] (we can give this a fancy name like 'doubly RELU') when you need to actually make predictions (as opposed to during training/testing), but if you're getting lots of such errors, the model might have more fundamental problems.
Even for a regression problem, it can be good (for optimisation) to use a sigmoid layer before the output (giving a prediction in the [0:1] range) followed by a denormalization (here if you think maximum age is 100, just multiply by 100)
This tip is explained in this fast.ai course.
I personally think these lessons are excellent.
You should use a sigmoid activation function, and then normalize the targets outputs to the [0, 1] range. This solves both issues of being positive and with a limit.
You can easily then denormalize the neural network outputs to get an output in the [0, 100] range.
Hi I was following the Machine Learning course by Andrew Ng.
I found that in regression problems, specially logistic regression they have used integer values for the features which could be plotted in a graph. But there are so many use cases where the feature values may not be integer.
Let's consider the follow example :
I want to build a model to predict if any particular person will take a leave today or not. From my historical data I may find the following features helpful to build the training set.
Name of the person, Day of the week, Number of Leaves left for him till now (which maybe a continuous decreasing variable), etc.
So here are the following questions based on above
How do I go about designing the training set for my logistic regression model.
In my training set, I find some variables are continuously decreasing (ex no of leaves left). Would that create any problem, because I know continuously increasing or decreasing variables are used in linear regression. Is that true ?
Any help is really appreciated. Thanks !
Well, there are a lot of missing information in your question, for example, it'll be very much clearer if you have provided all the features you have, but let me dare to throw some assumptions!
ML Modeling in classification always requires dealing with numerical inputs, and you can easily infer each of the unique input as an integer, especially the classes!
Now let me try to answer your questions:
How do I go about designing the training set for my logistic regression model.
How I see it, you have two options (not necessary both are practical, it's you who should decide according to the dataset you have and the problem), either you predict the probability of all employees in the company who will be off in a certain day according to the historical data you have (i.e. previous observations), in this case, each employee will represent a class (integer from 0 to the number of employees you want to include). Or you create a model for each employee, in this case the classes will be either off (i.e. Leave) or on (i.e. Present).
Example 1
I created a dataset example of 70 cases and 4 employees which looks like this:
Here each name is associated with the day and month they took as off with respect to how many Annual Leaves left for them!
The implementation (using Scikit-Learn) would be something like this (N.B date contains only day and month):
Now we can do something like this:
import math
import pandas as pd
import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import GridSearchCV, RepeatedStratifiedKFold
# read dataset example
df = pd.read_csv('leaves_dataset.csv')
# assign unique integer to every employee (i.e. a class label)
mapping = {'Jack': 0, 'Oliver': 1, 'Ruby': 2, 'Emily': 3}
df.replace(mapping, inplace=True)
y = np.array(df[['Name']]).reshape(-1)
X = np.array(df[['Leaves Left', 'Day', 'Month']])
# create the model
parameters = {'penalty': ['l1', 'l2'], 'C': [0.1, 0.5, 1.0, 10, 100, 1000]}
lr = LogisticRegression(random_state=0)
cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=2, random_state=0)
clf = GridSearchCV(lr, parameters, cv=cv)
clf.fit(X, y)
#print(clf.best_estimator_)
#print(clf.best_score_)
# Example: probability of all employees who have 10 days left today
# warning: date must be same format
prob = clf.best_estimator_.predict_proba([[10, 9, 11]])
print({'Jack': prob[0,0], 'Oliver': prob[0,1], 'Ruby': prob[0,2], 'Emily': prob[0,3]})
Result
{'Ruby': 0.27545, 'Oliver': 0.15032,
'Emily': 0.28201, 'Jack': 0.29219}
N.B
To make this relatively work you need a real big dataset!
Also this can be better than the second one if there are other informative features in the dataset (e.g. the health status of the employee at that day..etc).
The second option is to create a model for each employee, here the result would be more accurate and more reliable, however, it's almost a nightmare if you have too many employees!
For each employee, you collect all their leaves in the past years and concatenate them into one file, in this case you have to complete all days in the year, in other words: for every day that employee has never got it off, that day should be labeled as on (or numerically speaking 1) and for the days off they should be labeled as off (or numerically speaking 0).
Obviously, in this case, the classes will be 0 and 1 (i.e. on and off) for each employee's model!
For example, consider this dataset example for the particular employee Jack:
Example 2
Then you can do for example:
import pandas as pd
import numpy as np
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import GridSearchCV, RepeatedStratifiedKFold
# read dataset example
df = pd.read_csv('leaves_dataset2.csv')
# assign unique integer to every on and off (i.e. a class label)
mapping = {'off': 0, 'on': 1}
df.replace(mapping, inplace=True)
y = np.array(df[['Type']]).reshape(-1)
X = np.array(df[['Leaves Left', 'Day', 'Month']])
# create the model
parameters = {'penalty': ['l1', 'l2'], 'C': [0.1, 0.5, 1.0, 10, 100, 1000]}
lr = LogisticRegression(random_state=0)
cv = RepeatedStratifiedKFold(n_splits=10, n_repeats=2, random_state=0)
clf = GridSearchCV(lr, parameters, cv=cv)
clf.fit(X, y)
#print(clf.best_estimator_)
#print(clf.best_score_)
# Example: probability of the employee "Jack" who has 10 days left today
prob = clf.best_estimator_.predict_proba([[10, 9, 11]])
print({'Off': prob[0,0], 'On': prob[0,1]})
Result
{'On': 0.33348, 'Off': 0.66651}
N.B in this case you have to create a dataset for each employee + training especial model + filling all the days the never taken in the past years as off!
In my training set, I find some variables are continuously decreasing (ex no of leaves left). Would that create any problem,
because I know continuously increasing or decreasing variables are
used in linear regression. Is that true ?
Well, there is nothing preventing you from using contentious values as features (e.g. number of leaves) in Logistic Regression; actually it doesn't make any difference if it's used in Linear or Logistic Regression but I believe you got confused between the features and the response:
The thing is, discrete values should be used in the response of Logistic Regression and Continuous values should be used in the response of the Linear Regression (a.k.a dependent variable or y).
Data: When I have N rows of data like this: (x,y,z) where logically f(x,y)=z, that is z is dependent on x and y, like in my case (setting1, setting2 ,signal) . Different x's and y's can lead to the same z, but the z's wouldn't mean the same thing.
There are 30 unique setting1, 30 setting2 and 1 signal for each (setting1, setting2)-pairing, hence 900 signal values.
Data set: These [900,3] data points are considered 1 data set. I have many samples of these data sets.
I want to make a classification based on these data sets, but I need to flatten the data (make them all into one row). If I flatten it, I will duplicate all the setting values (setting1 and setting2) 30 times, i.e. I will have a row with 3x900 columns.
Question:
Is it correct to keep all the duplicate setting1,setting2 values in the data set? Or should I remove them and only include the unique values a single time?, i.e. have a row with 30 + 30 + 900 columns. I'm worried, that the logical dependency of the signal to the settings will be lost this way. Is this relevant? Or shouldn't I bother including the settings at all (e.g. due to correlations)?
If I understand correctly, you are training NN on a sample where each observation is [900,3].
You are flatning it and getting an input layer of 3*900.
Some of those values are a result of a function on others.
It is important which function, as if it is a liniar function, NN might not work:
From here:
"If inputs are linearly dependent then you are in effect introducing
the same variable as multiple inputs. By doing so you've introduced a
new problem for the network, finding the dependency so that the
duplicated inputs are treated as a single input and a single new
dimension in the data. For some dependencies, finding appropriate
weights for the duplicate inputs is not possible."
Also, if you add dependent variables you risk the NN being biased towards said variables.
E.g. If you are running LMS on [x1,x2,x3,average(x1,x2)] to predict y, you basically assign a higher weight to the x1 and x2 variables.
Unless you have a reason to believe that those weights should be higher, don't include their function.
I was not able to find any link to support, but my intuition is that you might want to decrease your input layer in addition to omitting the dependent values:
From professor A. Ng's ML Course I remember that the input should be the minimum amount of values that are 'reasonable' to make the prediction.
Reasonable is vague, but I understand it so: If you try to predict the price of a house include footage, area quality, distance from major hub, do not include average sun spot activity during the open home day even though you got that data.
I would remove the duplicates, I would also look for any other data that can be omitted, maybe run PCA over the full set of Nx[3,900].
I am using LSTM neural networks (stateful) for time series prediction.
I'm hoping that the stateful LSTM can capture the hidden patterns and make a satisfactory prediction (the physical law that cause the variation of the time series is not clear).
I have a time series X with a length of 1500 (actual observational data), and my purpose is to predict the future 100.
I suppose predict the next 10 will be more promising than predict the next 100 (is that right?).
So, I prepare the training data like this (always using 100 values to predict the next 10; x_n denotes the n-th element in X):
shape of trainX: [140, 100, 1]
shape of trainY: [140, 10, 1]
---
0: [x_0, x_1, ..., x_99] -> [x_100, x_101, ..., x_109]
1: [x_10, x_11, ..., x_109] -> [x_110, x_111, ..., x_119]
2: [x_20, x_21, ..., x_119] -> [x_120, x_121, ..., x_129]
...
139: [x_1390, x_1391, ..., x_1489] -> [x_1490, x_1491, ..., x_1499]
---
After the training, I want to use the model to predict the next 10 values [x_1500 - x_1509] with [x_1400 - x_1499], and then predict the next 10 values [x_1510 - x_1519] with [x_1410 - x_1509].
Is this the right way?
After a lot of reading of documents and examples, I can train a model and make the prediction, but the result seems not satisfactory.
To validate the method, I assume that the last 100 (x_1400 - x_1499) values are unknown, and remove them from trainX and trainY, then try to train a model and predict them. Lastly, compare the predicted values with the observed values.
Any suggestions or comments will be appreciated.
The time series looks like this:
Your question is really complexed. Before I will try to answer it - I'll share my doubts with you about is it sensible to use LSTM for your task. You want to use a really advanced model (LSTM are capable to learn really complexed patterns) to a time series which seems to be relatively easy. Moreover - you have a really small amout of data. To be honest - I would try to train simpler and easier methods first (like ARMA or ARIMA).
To answer your question - if your approach is good - it seems to be reasonable. Other reasonable methods are predicting all 100 steps or e.g. 50 steps twice. With 10 steps you might come across error cumulation - but still it might be a good method.
As I mentioned earlier - I would rather try easier ML method for this task but if you really want to use LSTM you may tackle this problem in a following way:
Define metaparameters like number of steps ahead you want to predict, the size of input fed to network.
Try to use e.g. grid search in order to find the best value of this metaparameters. Evaluate each setup using k-fold crossvalidation.
Retrain final model using the best metaparameter setup.
You have relatively small amount of data so you may easily find the best values of hyperparameters. This will also show you if your approach is good or not - simply check the results provided by the best solution.