I've created a neural network in tensorflow. This network is multilabel. Ergo: it tries to predict multiple output labels for one input set, in this case three. Currently I use this code to test how accurate my network is at predicting the three labels:
_, indices_1 = tf.nn.top_k(prediction, 3)
_, indices_2 = tf.nn.top_k(item_data, 3)
correct = tf.equal(indices_1, indices_2)
accuracy = tf.reduce_mean(tf.cast(correct, 'float'))
percentage = accuracy.eval({champion_data:input_data, item_data:output_data})
That code works fine. The problem is now that I'm trying to create code that tests if the top 3 items it finds in indices_1 are amongst the top 5 images in indices_2. I know tensorflow has an in_top_k() method, but as far as I know that doesn't accept multilabel. Currently I've been trying to compare them using a for loop:
_, indices_1 = tf.nn.top_k(prediction, 5)
_, indices_2 = tf.nn.top_k(item_data, 3)
indices_1 = tf.unpack(tf.transpose(indices_1, (1, 0)))
indices_2 = tf.unpack(tf.transpose(indices_2, (1, 0)))
correct = []
for element in indices_1:
for element_2 in indices_2:
if element == element_2:
correct.append(True)
else:
correct.append(False)
accuracy = tf.reduce_mean(tf.cast(correct, 'float'))
percentage = accuracy.eval({champion_data:input_data, item_data:output_data})
However, that doesn't work. The code runs but my accuracy is always 0.0.
So I have one of two questions:
1) Is there an easy replacement for in_top_k() that accepts multilabel classification that I can use instead of custom writing code?
2) If not 1: what am I doing wrong that results in me getting an accuracy of 0.0?
When you do
correct = tf.equal(indices_1, indices_2)
you are checking not just whether those two indices contain the same elements but whether they contain the same elements in the same positions. This doesn't sound like what you want.
The setdiff1d op will tell you which indices are in indices_1 but not in indices_2, which you can then use to count errors.
I think being too strict with the correctness check might be what is causing you to get a wrong result.
Related
When trying to create a table with the conditional random effects in r using the gtsummary function tbl_regression from a glmmTMB mixed effects negative-binomial zero-inflated model, I get duplicate random effects rows.
Example (using Mollie Brooks' Zero-Inflated GLMMs on Salamanders Dataset):
data(Salamanders)
head(Salamanders)
library(glmmTMB)
zinbm2 = glmmTMB(count~spp + mined +(1|site), zi=~spp + mined + (1|site), Salamanders, family=nbinom2)
zinbm2_table_cond <- tbl_regression(
zinbm2,
tidy_fun = function(...) broom.mixed::tidy(..., component = "cond"),
exponentiate = TRUE,
estimate_fun = purrr::partial(style_ratio, digits = 3),
pvalue_fun = purrr::partial(style_sigfig, digits = 3))
zinbm2_table_cond
Output:
Random Effects Output (cond)
When extracting the random effects from de zero-inflated part of the model I get the same problem.
Example:
zinbm2_table_zi <- tbl_regression(
zinbm2,
tidy_fun = function(...) broom.mixed::tidy(..., component = "zi"),
exponentiate = TRUE,
estimate_fun = purrr::partial(style_ratio, digits = 3),
pvalue_fun = purrr::partial(style_sigfig, digits = 3))
zinbm2_table_zi
Output:
Random Effects Output (zi)
The problem persists if I specify the effects argument in broom.mixed.
tidy_fun = function(...) broom.mixed::tidy(..., effects = "ran_pars", component = "cond"),
Looking at confidence intervals in both outputs it seems that somehow it is extracting random effects from both parts of the model and changing the estimate of the zero-inflated random effects (in 1st image; opposite in the 2nd image) to match the conditional part estimate while keeping the CI.
I am not knowledgeable enough to understand why this is happening. Since both rows have the same label I am having difficulty removing the wrong one.
Any tips on how to avoid this problem or a workaround to remove the undesired rows?
If you need more info, let me know.
Thank you in advance.
PS: Output images were changed to link due to insufficient reputation.
I am trying to use k-medoids to cluster some trajectory data I am working with (multiple points along the trajectory of an aircraft). I want to cluster these into a set number of clusters (as I know how many types of paths there should be).
I have found that k-medoids is implemented inside the pyclustering package, and am trying to use that. I am technically able to get it to cluster, but I do not know how to control the number of clusters. I originally thought it was directly tied to the number of elements inside what I called initial_medoids, but experimentation shows that it is more complicated than this. My relevant code snippet is below.
Note that D holds a list of lists. Each list corresponds to a single trajectory.
def hausdorff( u, v):
d = max(directed_hausdorff(u, v)[0], directed_hausdorff(v, u)[0])
return d
traj_count = len(traj_lst)
D = np.zeros((traj_count, traj_count))
for i in range(traj_count):
for j in range(i + 1, traj_count):
distance = hausdorff(traj_lst[i], traj_lst[j])
D[i, j] = distance
D[j, i] = distance
from pyclustering.cluster.kmedoids import kmedoids
initial_medoids = [104, 345, 123, 1]
kmedoids_instance = kmedoids(traj_lst, initial_medoids)
kmedoids_instance.process()
cluster_lst = kmedoids_instance.get_clusters()[0]
num_clusters = len(np.unique(cluster_lst))
print('There were %i clusters found' %num_clusters)
I have a total of 1900 trajectories, and the above-code finds 1424 clusters. I had expected that I could control the number of clusters through the length of initial_medoids, as I did not see any option to input the number of clusters into the program, but this seems unrelated. Could anyone guide me as to the mistake I am making? How do I choose the number of clusters?
In case of requirement to obtain clusters you need to call get_clusters():
cluster_lst = kmedoids_instance.get_clusters()
Not get_clusters()[0] (in this case it is a list of object indexes in the first cluster):
cluster_lst = kmedoids_instance.get_clusters()[0]
And that is correct, you can control amount of clusters by initial_medoids.
It is true you can control the number of cluster, which correspond to the length of initial_medoids.
The documentation is not clear about this. The get__clusters function "Returns list of medoids of allocated clusters represented by indexes from the input data". so, this function does not return the cluster labels. It returns the index of rows in your original (input) data.
Please check the shape of cluster_lst in your example, using .get_clusters() and not .get_clusters()[0] as annoviko suggested. In your case, this shape should be (4,). So, you have a list of four elements (clusters), each containing the index or rows in your original data.
To get, for example, data from the first cluster, use:
kmedoids_instance = kmedoids(traj_lst, initial_medoids)
kmedoids_instance.process()
cluster_lst = kmedoids_instance.get_clusters()
traj_lst_first_cluster = traj_lst[cluster_lst[0]]
So I have a data set where I do some transformations and the last step is to filter out rows that have a 0 in a column called frequency. The code that does the filtering is super simple:
def filter_rows(self, name: str = None, frequency_col: str = 'frequency', threshold: int = 1):
df = getattr(self, name)
df = df.where(df[frequency_col] >= threshold)
setattr(self, name, df)
return self
The problem is a very strange behavior where if I put a rather high threshold like 10, it works fine, filtering out all the rows below 10. But if I make the threshold just 1, it does not remove the 0s! Here is an example of the former (threshold=10):
{"user":"XY1677KBTzDX7EXnf-XRAYW4ZB_vmiNvav7hL42BOhlcxZ8FQ","domain":"3a899ebbaa182778d87d","frequency":12}
{"user":"lhoAWb9U9SXqscEoQQo9JqtZo39nutq3NgrJjba38B10pDkI","domain":"3a899ebbaa182778d87d","frequency":9}
{"user":"aRXbwY0HcOoRT302M8PCnzOQx9bOhDG9Z_fSUq17qtLt6q6FI","domain":"33bd29288f507256d4b2","frequency":23}
{"user":"RhfrV_ngDpJex7LzEhtgmWk","domain":"390b4f317c40ac486d63","frequency":14}
{"user":"qZqqsNSNko1V9eYhJB3lPmPp0p5bKSq0","domain":"390b4f317c40ac486d63","frequency":11}
{"user":"gsmP6RG13azQRmQ-RxcN4MWGLxcx0Grs","domain":"f4765996305ccdfa9650","frequency":10}
{"user":"jpYTnYjVkZ0aVexb_L3ZqnM86W8fr082HwLliWWiqhnKY5A96zwWZKNxC","domain":"f4765996305ccdfa9650","frequency":15}
{"user":"Tlgyxk_rJF6uE8cLM2sArPRxiOOpnLwQo2s","domain":"f89838b928d5070c3bc3","frequency":17}
{"user":"qHu7fpnz2lrBGFltj98knzzbwWDfU","domain":"f89838b928d5070c3bc3","frequency":11}
{"user":"k0tU5QZjRkBwqkKvMIDWd565YYGHfg","domain":"f89838b928d5070c3bc3","frequency":17}
And now here is some of the data with threshold=1:
{"user":"KuhSEPFKACJdNyMBBD2i6ul0Nc_b72J4","domain":"d69cb6f62b885fec9b7d","frequency":0}
{"user":"EP1LomZ3qAMV3YtduC20","domain":"d69cb6f62b885fec9b7d","frequency":0}
{"user":"UxulBfshmCro-srE3Cs5znxO5tnVfc0_yFps","domain":"d69cb6f62b885fec9b7d","frequency":1}
{"user":"v2OX7UyvMVnWlDeDyYC8Opk-va_i8AwxZEsxbk","domain":"d69cb6f62b885fec9b7d","frequency":0}
{"user":"4hu1uE2ucAYZIrNLeOY2y9JMaArFZGRqjgKzlKenC5-GfxDJQQbLcXNSzj","domain":"68b588cedbc66945c442","frequency":0}
{"user":"5rFMWm_A-7N1E9T289iZ65TIR_JG_OnZpJ-g","domain":"68b588cedbc66945c442","frequency":1}
{"user":"RLqoxFMZ7Si3CTPN1AnI4hj6zpwMCJI","domain":"68b588cedbc66945c442","frequency":1}
{"user":"wolq9L0592MGRfV_M-FxJ5Wc8UUirjqjMdaMDrI","domain":"68b588cedbc66945c442","frequency":0}
{"user":"9spTLehI2w0fHcxyvaxIfo","domain":"68b588cedbc66945c442","frequency":1}
I should note that before this step I perform some other transformations, and I've noticed weird behaviors in Spark in the past sometimes doing very simple things like this after a join or a union can give very strange results where eventually the only solution is to write out the data and read it back in again and do the operation in a completely separate script. I hope there is a better solution than this!
I am using a Keras deep autoencoder to reproduce my sparse matrix of [360, 6860] dimension. Each row is the count of trigrams for a protein sequence. The matrix has 2 classes of proteins, but I want the network to be ignorant of that initially, that is why I am using an autoencoder. I am following the keras blog autoencoder tutorial for this.
This is my code-
# this is the size of our encoded representations
encoding_dim = 32
input_img = Input(shape=(6860,))
encoded = Dense(128, activation='relu', activity_regularizer=regularizers.activity_l1(10e-5))(input_img)
encoded = Dense(64, activation='relu')(encoded)
encoded = Dense(32, activation='relu')(encoded)
decoded = Dense(64, activation='relu')(encoded)
decoded = Dense(128, activation='relu')(decoded)
decoded = Dense(6860, activation='sigmoid')(decoded)
autoencoder = Model(input=input_img, output=decoded)
# this model maps an input to its encoded representation
encoder = Model(input=input_img, output=encoded)
# create a placeholder for an encoded (32-dimensional) input
encoded_input_1 = Input(shape=(32,))
encoded_input_2 = Input(shape=(64,))
encoded_input_3 = Input(shape=(128,))
# retrieve the last layer of the autoencoder model
decoder_layer_1 = autoencoder.layers[-3]
decoder_layer_2 = autoencoder.layers[-2]
decoder_layer_3 = autoencoder.layers[-1]
# create the decoder model
decoder_1 = Model(input = encoded_input_1, output = decoder_layer_1(encoded_input_1))
decoder_2 = Model(input = encoded_input_2, output = decoder_layer_2(encoded_input_2))
decoder_3 = Model(input = encoded_input_3, output = decoder_layer_3(encoded_input_3))
autoencoder.compile(optimizer='adadelta', loss='binary_crossentropy')
autoencoder.fit(x_train, x_train,
nb_epoch= 100,
batch_size=40,
shuffle=True,
validation_data=(x_test, x_test))
My validation set dimension is [80, 6860]. The problem is if I use the decoder to predict from the test set, my predictions are really off. For example if I predict with the following code-
# encode and decode some digits
# note that we take them from the *test* set
encoded_imgs = encoder.predict(x_test)
decoded_imgs = decoder_1.predict(encoded_imgs)
decoded_imgs = decoder_2.predict(decoded_imgs)
decoded_imgs = decoder_3.predict(decoded_imgs)
print x_test[3, np.where(x_test[3, :] != 0)[0]]
print (decoded_imgs[3, np.where(x_test[3, :] != 0)[0]])
a single row of my test set where the values are not zero are-
[ 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 2. 1. 1. 1. 1.
1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1.
1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1. 1.]
for the same row, the autoencoder's prediction of the same indices are-
[ 0.04615583 0.04613763 0.10268984 0.00286385 0.0030572 0.02551027
0.00552908 0.09686473 0.02554915 0.0082816 0.02254158 0.01127195
0.00305908 0.17113154 0.01140419 0.03370495 0.00515486 0.02614204
0.00558715 0.02835727 0.0029659 0.01425297 0.00834536 0.04502939
0.02260707 0.01131396 0.00561662 0.01131314 0.00493734 0.00265232
0.0056083 0.01724379 0.06099484 0.03738695 0.01128869 0.01995548
0.00562622 0.00556281 0.01732991 0.03142899 0.05339266 0.04778111
0.00292415 0.02264618 0.01419865 0.00550648 0.00836777 0.01139715]
Now, first I thought, maybe I can use some kind of thresholding to get the 1's from these values. But it seems they are pretty random. For a single row, for the first 50 zero values for my test set, my autoencoder predicts-
[ 0.14251608 0.00118295 0.00118732 0.00304095 0.031255 0.00108441
0.0201351 0.00853934 0.00558488 0.00281343 0.00296877 0.00109651
0.01129742 0.00827519 0.0170884 0.01417614 0.01714166 0.00549215
0.00099755 0.00558552 0.00829634 0.01988331 0.00092845 0.00294271
0.01429107 0.01137067 0.01137967 0.01121876 0.00491931 0.00562285
0.0055124 0.01720702 0.0142925 0.00553411 0.00551252 0.00281541
0.01145663 0.002876 0.00555185 0.00525392 0.01421779 0.00273949
0.01698892 0.02529835 0.0112521 0.01130333 0.00554186 0.00291986
0.00554437 0.01144382]
How can I improve the predictions? What am I doing wrong here? I must say that the data is hugely sparse. If you want you can download the toy data from here. Please, let me know if you have any questions.
One of the most important reasons is probably your training data size is just too small. You have a fully connected network and thus with 7 layers (including input and output) the number of parameters are just huge, close to 1.8M. You only have 360 training samples. So basically the parameters are untrained.
You can improve your work in two ways. One is of course to get more training data. The second is to follow the CNN example in the later part of the tutorial. CNN has been popular since it can greatly reduce the number of parameters.
I'm using RandomForest for classification, and I got an unbalanced dataset, as: 5830-no, 1006-yes. I try to balance my dataset with class_weight and sample_weight, but I can`t.
My code is:
X_train,X_test,y_train,y_test = train_test_split(arrX,y,test_size=0.25)
cw='auto'
clf=RandomForestClassifier(class_weight=cw)
param_grid = { 'n_estimators': [10,50,100,200,300],'max_features': ['auto', 'sqrt', 'log2']}
sw = np.array([1 if i == 0 else 8 for i in y_train])
CV_clf = GridSearchCV(estimator=clf, param_grid=param_grid, cv= 10,fit_params={'sample_weight': sw})
But I don't get any improvement on my ratios TPR, FPR, ROC when using class_weight and sample_weight.
Why? Am I doing anything wrong?
Nevertheless, if I use the function called balanced_subsample, my ratios obtain a great improvement:
def balanced_subsample(x,y,subsample_size):
class_xs = []
min_elems = None
for yi in np.unique(y):
elems = x[(y == yi)]
class_xs.append((yi, elems))
if min_elems == None or elems.shape[0] < min_elems:
min_elems = elems.shape[0]
use_elems = min_elems
if subsample_size < 1:
use_elems = int(min_elems*subsample_size)
xs = []
ys = []
for ci,this_xs in class_xs:
if len(this_xs) > use_elems:
np.random.shuffle(this_xs)
x_ = this_xs[:use_elems]
y_ = np.empty(use_elems)
y_.fill(ci)
xs.append(x_)
ys.append(y_)
xs = np.concatenate(xs)
ys = np.concatenate(ys)
return xs,ys
My new code is:
X_train_subsampled,y_train_subsampled=balanced_subsample(arrX,y,0.5)
X_train,X_test,y_train,y_test = train_test_split(X_train_subsampled,y_train_subsampled,test_size=0.25)
cw='auto'
clf=RandomForestClassifier(class_weight=cw)
param_grid = { 'n_estimators': [10,50,100,200,300],'max_features': ['auto', 'sqrt', 'log2']}
sw = np.array([1 if i == 0 else 8 for i in y_train])
CV_clf = GridSearchCV(estimator=clf, param_grid=param_grid, cv= 10,fit_params={'sample_weight': sw})
This is not a full answer yet, but hopefully it'll help get there.
First some general remarks:
To debug this kind of issue it is often useful to have a deterministic behavior. You can pass the random_state attribute to RandomForestClassifier and various scikit-learn objects that have inherent randomness to get the same result on every run. You'll also need:
import numpy as np
np.random.seed()
import random
random.seed()
for your balanced_subsample function to behave the same way on every run.
Don't grid search on n_estimators: more trees is always better in a random forest.
Note that sample_weight and class_weight have a similar objective: actual sample weights will be sample_weight * weights inferred from class_weight.
Could you try:
Using subsample=1 in your balanced_subsample function. Unless there's a particular reason not to do so we're better off comparing the results on similar number of samples.
Using your subsampling strategy with class_weight and sample_weight both set to None.
EDIT: Reading your comment again I realize your results are not so surprising!
You get a better (higher) TPR but a worse (higher) FPR.
It just means your classifier tries hard to get the samples from class 1 right, and thus makes more false positives (while also getting more of those right of course!).
You will see this trend continue if you keep increasing the class/sample weights in the same direction.
There is a imbalanced-learn API that helps with oversampling/undersampling data that might be useful in this situation. You can pass your training set into one of the methods and it will output the oversampled data for you. See simple example below
from imblearn.over_sampling import RandomOverSampler
ros = RandomOverSampler(random_state=1)
x_oversampled, y_oversampled = ros.fit_sample(orig_x_data, orig_y_data)
Here it the link to the API: http://contrib.scikit-learn.org/imbalanced-learn/api.html
Hope this helps!