I am building a Logistic Regression platform in PHP. The following code snippet works fine when there is only one feature inside the data frame. For example a CSV file like this:
"sample","language"
"Hello, how are you?","english",
"Je voudrais une boîte de chocolats.","french"
...
However, when I try to train the AI with 2 features based on the titanic survival rate (hypothesis: Does the amount of siblings and spouses effect the survival rate) with a data frame like this:
"SibSp","Parch","Survived",
"1", "1", "1",
"3", "3", "1",
"4", "1", "0"
...
I am getting this error:
Phpml\\Exception\\InvalidArgumentException Size of given arrays does not match
My code snippet looks like this, $request->features
holds the amount of features this data frame has since features +1
will hold the actual outcome (1 = survived, 0 = died):
$dataset = new CsvDataset($file, (int) $request->features);
$vectorizer = new TokenCountVectorizer(new WordTokenizer());
$tfIdfTransformer = new TfIdfTransformer();
$samples = [];
for($i = 0; $i <= $request->features -1; $i++):
foreach ($dataset->getSamples() as $sample):
$samples[$i][] = $sample[$i];
endforeach;
endfor;
for($i = 0; $i <= count($samples) -1; $i++):
$vectorizer->fit($samples[$i]);
$vectorizer->transform($samples[$i]);
$tfIdfTransformer->fit($samples[$i]);
$tfIdfTransformer->transform($samples[$i]);
endfor;
$dataset = new ArrayDataset($samples, $dataset->getTargets()); # This throws the error
I am using PHP-AI/PHP-ML and here is an example of how the AI works with the data frame with only 1 feature provided by the framework.
I understand the error, $dataset->getTargets()
only holds 1 array, where as $samples
holds 2 arrays. However, this has got me stumped since that is how it should be (in theory).
I am storing the classifier (or trained AI) as a serialised object inside my database once it has been trained to remember its trained state. Everything works fine when I only use a data frame with one feature. Does anyone have experience using PHP-AI within the PHP-ML library that can help?
How can I increase the amount of features inside PHP-AI?
Update to show what values my arrays hold:
$samples
looks like this (array of siblings, array of spouses):
array ( 0 => array ( 0 => array ( ), 1 => array ( ), 2 => array ( ), 3 => array ( ), 4 => array ( ), 5 => array ( ), 6 => array ( ), 7 => array ( ), ), 1 => array ( 0 => array ( ), 1 => array ( ), 2 => array ( ), 3 => array ( ), 4 => array ( ), 5 => array ( ), 6 => array ( ), 7 => array ( ), ), )
$dataset->getTargets()
looks like this (survived or died):
array ( 0 => '1', 1 => '1', 2 => '0', 3 => '1', 4 => '0', 5 => '0', 6 => '1', 7 => '1', )
I believe that the $samples
array should be 1 array holding child arrays of [SibSp, Spous]. I cannot think how to re-organise the array to be like this.
After fiddling around with the code and researching the error and how to get around it - I realised that the $samples
data should be expressed as
Array [ 0 => [SibSp, Spous], 1 => [SibSp, Spous], ... ]
So by re-fiddling the data like so:
$result = [];
foreach($samples as $arr) {
foreach($arr as $k => $v) {
$result[$k][] = $v;
}
}
I can achieve this desired outcome. I still had to push the samples into the vectorizer as $sample
but the final Dataset had to be re-fiddled:
for($i = 0; $i <= count($samples) -1; $i++):
$vectorizer->fit($samples[$i]);
$vectorizer->transform($samples[$i]);
$tfIdfTransformer->fit($samples[$i]);
$tfIdfTransformer->transform($samples[$i]);
endfor;
$dataset = new ArrayDataset($result, $dataset->getTargets());
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.