scicloj.ml.dataset — scicloj/scicloj.ml 0.1.0-beta8

This namespace contains functions which operate on a dataset and mostly return a dataset.

The namespaces scicloj.ml.metamorph and scicloj.ml.dataset contain functions with the same name. But they operate on either a context map (ns metamorph) or on a dataset (ns dataset)

The functions in tis namespace are re-exported from:

• tabecloth.api
• tech.v3.dataset.modelling
• tech.v3.dataset.column-filters

->array clj

(->array ds colname)

(->array ds colname datatype)

Convert numerical column(s) to java array

add-column clj

(add-column ds column-name column)

(add-column ds column-name column size-strategy)

Add or update (modify) column under column-name.

column can be sequence of values or generator function (which gets ds as input).

add-columns clj

(add-columns ds columns-map)

(add-columns ds columns-map size-strategy)

Add or updade (modify) columns defined in columns-map (mapping: name -> column)

add-or-replace-column clj

(add-or-replace-column ds column-name column)

(add-or-replace-column ds column-name column size-strategy)

add-or-replace-columns clj

(add-or-replace-columns ds columns-map)

(add-or-replace-columns ds columns-map size-strategy)

aggregate clj

(aggregate ds aggregator)

(aggregate ds
           aggregator
           {:keys [default-column-name-prefix ungroup? parallel?]
            :or {default-column-name-prefix "summary" ungroup? true}
            :as options})

Aggregate dataset by providing:

• aggregation function
• map with column names and functions
• sequence of aggregation functions

Aggregation functions can return:

• single value
• seq of values
• map of values with column names

aggregate-columns clj

(aggregate-columns ds columns-selector column-aggregators)

(aggregate-columns ds columns-selector column-aggregators options)

Aggregates each column separately

anti-join clj

(anti-join ds-left ds-right columns-selector)

(anti-join ds-left ds-right columns-selector options)

append clj

(append ds & datasets)

as-regular-dataset clj

(as-regular-dataset ds)

Remove grouping tag

asof-join clj

(asof-join ds-left ds-right colname)

(asof-join ds-left ds-right colname options)

bind clj

(bind ds & datasets)

boolean clj

(boolean dataset)

Return a dataset containing only the boolean columns.

by-rank clj

(by-rank ds columns-selector rank-predicate)

(by-rank ds
         columns-selector
         rank-predicate
         {:keys [desc? ties] :or {desc? true ties :dense}})

Select rows using rank on a column, ties are resolved using :dense method.

See R docs. Rank uses 0 based indexing.

Possible :ties strategies: :average, :first, :last, :random, :min, :max, :dense. :dense is the same as in data.table::frank from R

:desc? set to true (default) order descending before calculating rank

categorical clj

(categorical dataset)

Return a dataset containing only the categorical columns.

categorical->number clj

(categorical->number dataset filter-fn-or-ds)

(categorical->number dataset filter-fn-or-ds table-args)

(categorical->number dataset filter-fn-or-ds table-args result-datatype)

Convert columns into a discrete , numeric representation See tech.v3.dataset.categorical/fit-categorical-map.

categorical->one-hot clj

(categorical->one-hot dataset filter-fn-or-ds)

(categorical->one-hot dataset filter-fn-or-ds table-args)

(categorical->one-hot dataset filter-fn-or-ds table-args result-datatype)

Convert string columns to numeric columns. See tech.v3.dataset.categorical/fit-one-hot

CategoricalMap clj

clone clj

(clone item)

Clone an object. Can clone anything convertible to a reader.

column clj

(column dataset colname)

column-count clj

(column-count dataset)

column-filter clj

(column-filter dataset filter-fn)

Return a dataset with only the columns for which the filter function returns a truthy value.

column-names clj

(column-names ds)

(column-names ds columns-selector)

(column-names ds columns-selector meta-field)

column-values->categorical clj

(column-values->categorical dataset src-column)

Given a column encoded via either string->number or one-hot, reverse map to the a sequence of the original string column values. In the case of one-hot mappings, src-column must be the original column name before the one-hot map

columns clj

(columns ds)

(columns ds result-type)

concat clj

(concat dataset & datasets)

concat-copying clj

(concat-copying dataset & datasets)

convert-types clj

(convert-types ds coltype-map-or-columns-selector)

(convert-types ds columns-selector new-types)

Convert type of the column to the other type.

create-categorical-map clj

(create-categorical-map lookup-table src-colname result-datatype)

dataset clj

(dataset)

(dataset data)

(dataset data
         {:keys [single-value-column-name column-names layout dataset-name]
          :or {single-value-column-name :$value layout :as-rows}
          :as options})

Create dataset.

Dataset can be created from:

• single value
• map of values and/or sequences
• sequence of maps
• sequence of columns
• file or url

dataset->categorical-maps clj

(dataset->categorical-maps dataset)

Given a dataset, return a map of column names to categorical label maps. This aids in inverting all of the label maps in a dataset. The source column name is src-column.

dataset->categorical-xforms clj

(dataset->categorical-xforms ds)

Given a dataset, return a map of column-name->xform information.

dataset->one-hot-maps clj

(dataset->one-hot-maps dataset)

Given a dataset, return a sequence of applied on-hot transformations.

dataset->str clj

(dataset->str ds)

(dataset->str ds options)

Convert a dataset to a string. Prints a single line header and then calls dataset-data->str.

For options documentation see dataset-data->str.

dataset-name clj

(dataset-name dataset)

dataset? clj

(dataset? ds)

Is ds a dataset type?

datetime clj

(datetime dataset)

Return a dataset containing only the datetime columns.

difference clj

(difference ds-left ds-right)

(difference ds-left ds-right options)

drop clj

(drop ds columns-selector rows-selector)

Drop columns and rows.

drop-columns clj

(drop-columns ds)

(drop-columns ds columns-selector)

(drop-columns ds columns-selector meta-field)

Drop columns by (returns dataset):

• name
• sequence of names
• map of names with new names (rename)
• function which filter names (via column metadata)

drop-missing clj

(drop-missing ds)

(drop-missing ds columns-selector)

Drop rows with missing values

columns-selector selects columns to look at missing values

drop-rows clj

(drop-rows ds)

(drop-rows ds rows-selector)

(drop-rows ds rows-selector {:keys [select-keys pre result-type parallel?]})

Drop rows using:

• row id
• seq of row ids
• seq of true/false
• fn with predicate

empty-ds? clj

(empty-ds? ds)

feature clj

(feature dataset)

Return a dataset container only the columns which have not been marked as inference columns.

feature-ecount clj

(feature-ecount dataset)

Number of feature columns. Feature columns are columns that are not inference targets.

fill-range-replace clj

(fill-range-replace ds colname max-span)

(fill-range-replace ds colname max-span missing-strategy)

(fill-range-replace ds colname max-span missing-strategy missing-value)

first clj

(first ds)

fit-categorical-map clj

(fit-categorical-map dataset colname & [table-args res-dtype])

Given a column, map it into an numeric space via a discrete map of values to integers. This fits the categorical transformation onto the column and returns the transformation.

If table-args is not given, the distinct column values will be mapped into 0..x without any specific order.

'table-args` allows to specify the precise mapping as a sequence of pairs of [val idx] or as a sorted seq of values.

fit-one-hot clj

(fit-one-hot dataset colname & [table-args res-dtype])

Fit a one hot transformation to a column. Returns a reusable transformation. Maps each unique value to a column with 1 every time the value appears in the original column and 0 otherwise.

fold-by clj

(fold-by ds columns-selector)

(fold-by ds columns-selector folding-function)

full-join clj

(full-join ds-left ds-right columns-selector)

(full-join ds-left ds-right columns-selector options)

group-by clj

(group-by ds grouping-selector)

(group-by ds
          grouping-selector
          {:keys [select-keys result-type]
           :or {result-type :as-dataset select-keys :all}
           :as options})

Group dataset by:

• column name
• list of columns
• map of keys and row indexes
• function getting map of values

Options are:

• select-keys - when grouping is done by function, you can limit fields to a select-keys seq.
• result-type - return results as dataset (:as-dataset, default) or as map of datasets (:as-map) or as map of row indexes (:as-indexes) or as sequence of (sub)datasets
• other parameters which are passed to dataset fn

When dataset is returned, meta contains :grouped? set to true. Columns in dataset:

• name - group name
• group-id - id of the group (int)
• data - group as dataset

grouped? clj

(grouped? ds)

Is dataset represents grouped dataset (result of group-by)?

groups->map clj

(groups->map ds)

Convert grouped dataset to the map of groups

groups->seq clj

(groups->seq ds)

has-column? clj

(has-column? dataset column-name)

head clj

(head ds)

(head ds n)

inference-column? clj

(inference-column? col)

inference-target-column-names clj

(inference-target-column-names ds)

Return the names of the columns that are inference targets.

inference-target-ds clj

(inference-target-ds dataset)

Given a dataset return reverse-mapped inference target columns or nil in the case where there are no inference targets.

inference-target-label-inverse-map clj

(inference-target-label-inverse-map dataset & [label-columns])

Given options generated during ETL operations and annotated with :label-columns sequence container 1 label column, generate a reverse map that maps from a dataset value back to the label that generated that value.

inference-target-label-map clj

(inference-target-label-map dataset & [label-columns])

info clj

(info ds)

(info ds result-type)

inner-join clj

(inner-join ds-left ds-right columns-selector)

(inner-join ds-left ds-right columns-selector options)

intersect clj

(intersect ds-left ds-right)

(intersect ds-left ds-right options)

intersection clj

(intersection lhs-ds rhs-ds)

Return only columns for rhs for which an equivalently named column exists in lhs.

invert-categorical-map clj

(invert-categorical-map dataset {:keys [src-column lookup-table]})

Invert a categorical map returning the column to the original set of values.

invert-one-hot-map clj

(invert-one-hot-map dataset {:keys [one-hot-table src-column]})

Invert a one-hot transformation removing the one-hot columns and adding back the original column.

join-columns clj

(join-columns ds target-column columns-selector)

(join-columns ds
              target-column
              columns-selector
              {:keys [separator missing-subst drop-columns? result-type
                      parallel?]
               :or {separator "-" drop-columns? true result-type :string}})

k-fold-datasets clj

(k-fold-datasets dataset k)

(k-fold-datasets dataset k options)

Given 1 dataset, prepary K datasets using the k-fold algorithm. Randomize dataset defaults to true which will realize the entire dataset so use with care if you have large datasets.

Returns a sequence of {:test-ds :train-ds}

Options:

• :randomize-dataset? - When true, shuffle the dataset. In that case 'seed' may be provided. Defaults to true.
• :seed - when :randomize-dataset? is true then this can either be an implementation of java.util.Random or an integer seed which will be used to construct java.util.Random.

labels clj

(labels dataset)

Return the labels. The labels sequence is the reverse mapped inference column. This returns a single column of data or errors out.

last clj

(last ds)

left-join clj

(left-join ds-left ds-right columns-selector)

(left-join ds-left ds-right columns-selector options)

map-columns clj

(map-columns ds column-name map-fn)

(map-columns ds column-name columns-selector map-fn)

(map-columns ds column-name new-type columns-selector map-fn)

mark-as-group clj

(mark-as-group ds)

Add grouping tag

metadata-filter clj

(metadata-filter dataset filter-fn)

Return a dataset with only the columns for which, given the column metadata, the filter function returns a truthy value.

missing clj

(missing dataset)

Return a dataset with only columns have have missing values

model-type clj

(model-type dataset & [column-name-seq])

Check the label column after dataset processing. Return either :regression :classification

no-missing clj

(no-missing dataset)

Return a dataset with only columns that have no missing values.

num-inference-classes clj

(num-inference-classes dataset)

Given a dataset and correctly built options from pipeline operations, return the number of classes used for the label. Error if not classification dataset.

numeric clj

(numeric dataset)

Return a dataset containing only the numeric columns.

of-datatype clj

(of-datatype dataset datatype)

Return a dataset containing only the columns of a specific datatype.

OneHotMap clj

order-by clj

(order-by ds columns-or-fn)

(order-by ds columns-or-fn comparators)

(order-by ds columns-or-fn comparators {:keys [parallel?]})

Order dataset by:

• column name
• columns (as sequence of names)
• key-fn
• sequence of columns / key-fn Additionally you can ask the order by:
• :asc
• :desc
• custom comparator function

pivot->longer clj

(pivot->longer ds)

(pivot->longer ds columns-selector)

(pivot->longer
  ds
  columns-selector
  {:keys [target-columns value-column-name splitter drop-missing? datatypes]
   :or {target-columns :$column value-column-name :$value drop-missing? true}})

tidyr pivot_longer api

pivot->wider clj

(pivot->wider ds columns-selector value-columns)

(pivot->wider
  ds
  columns-selector
  value-columns
  {:keys [fold-fn concat-columns-with concat-value-with drop-missing?]
   :or {concat-columns-with "_" concat-value-with "-" drop-missing? true}})

prediction clj

(prediction dataset)

Return the columns of the dataset marked as predictions.

print-dataset clj

(print-dataset ds)

(print-dataset ds options)

probability-distribution clj

(probability-distribution dataset)

Return the columns of the dataset that comprise the probability distribution after classification.

probability-distributions->label-column clj

(probability-distributions->label-column prob-ds dst-colname)

Given a dataset that has columns in which the column names describe labels and the rows describe a probability distribution, create a label column by taking the max value in each row and assign column that row value.

process-group-data clj

(process-group-data ds f)

(process-group-data ds f parallel?)

rand-nth clj

(rand-nth ds)

(rand-nth ds {:keys [seed]})

random clj

(random ds)

(random ds n)

(random ds n {:keys [repeat? seed] :or {repeat? true}})

read-nippy clj

(read-nippy filename)

rename-columns clj

(rename-columns ds columns-mapping)

(rename-columns ds columns-selector columns-map-fn)

Rename columns with provided old -> new name map

reorder-columns clj

(reorder-columns ds columns-selector & columns-selectors)

Reorder columns using column selector(s). When column names are incomplete, the missing will be attached at the end.

replace-missing clj

(replace-missing ds)

(replace-missing ds strategy)

(replace-missing ds columns-selector strategy)

(replace-missing ds columns-selector strategy value)

reverse-map-categorical-xforms clj

(reverse-map-categorical-xforms dataset)

Given a dataset where we have converted columns from a categorical representation to either a numeric reprsentation or a one-hot representation, reverse map back to the original dataset given the reverse mapping of label->number in the column's metadata.

right-join clj

(right-join ds-left ds-right columns-selector)

(right-join ds-left ds-right columns-selector options)

row-count clj

(row-count dataset-or-col)

rows clj

(rows ds)

(rows ds result-type)

select clj

(select ds columns-selector rows-selector)

Select columns and rows.

select-columns clj

(select-columns ds)

(select-columns ds columns-selector)

(select-columns ds columns-selector meta-field)

Select columns by (returns dataset):

• name
• sequence of names
• map of names with new names (rename)
• function which filter names (via column metadata)

select-missing clj

(select-missing ds)

(select-missing ds columns-selector)

Select rows with missing values

columns-selector selects columns to look at missing values

select-rows clj

(select-rows ds)

(select-rows ds rows-selector)

(select-rows ds rows-selector {:keys [select-keys pre result-type parallel?]})

Select rows using:

• row id
• seq of row ids
• seq of true/false
• fn with predicate

semi-join clj

(semi-join ds-left ds-right columns-selector)

(semi-join ds-left ds-right columns-selector options)

separate-column clj

(separate-column ds column separator)

(separate-column ds column target-columns separator)

(separate-column ds
                 column
                 target-columns
                 separator
                 {:keys [missing-subst drop-column? parallel?]
                  :or {missing-subst ""}})

set-dataset-name clj

(set-dataset-name dataset ds-name)

set-inference-target clj

(set-inference-target dataset target-name-or-target-name-seq)

Set the inference target on the column. This sets the :column-type member of the column metadata to :inference-target?.

shape clj

(shape ds)

Returns shape of the dataset [rows, cols]

shuffle clj

(shuffle ds)

(shuffle ds {:keys [seed]})

split clj

(split ds)

(split ds split-type)

(split ds
       split-type
       {:keys [seed parallel? shuffle?] :or {shuffle? true} :as opts})

Split given dataset into 2 or more (holdout) splits

As the result two new columns are added:

• :$split-name - with subgroup name
• :$split-id - fold id/repetition id

split-type can be one of the following:

• :kfold - k-fold strategy, :k defines number of folds (defaults to 5), produces k splits
• :bootstrap - :ratio defines ratio of observations put into result (defaults to 1.0), produces 1 split
• :holdout - split into two parts with given ratio (defaults to 2/3), produces 1 split
• :loo - leave one out, produces the same number of splits as number of observations

:holdout can accept also probabilites or ratios and can split to more than 2 subdatasets

Additionally you can provide:

• :seed - for random number generator
• :repeats - repeat procedure :repeats times
• :partition-selector - same as in group-by for stratified splitting to reflect dataset structure in splits.
• :split-names names of subdatasets different than default, ie. [:train :test :split-2 ...]
• :split-col-name - a column where name of split is stored, either :train or :test values (default: :$split-name)
• :split-id-col-name - a column where id of the train/test pair is stored (default: :$split-id)

Rows are shuffled before splitting.

In case of grouped dataset each group is processed separately.

See more

split->seq clj

(split->seq ds)

(split->seq ds split-type)

(split->seq ds
            split-type
            {:keys [split-col-name split-id-col-name]
             :or {split-col-name :$split-name split-id-col-name :$split-id}
             :as opts})

Returns split as a sequence of train/test datasets or map of sequences (grouped dataset)

string clj

(string dataset)

Return a dataset containing only the string columns.

tail clj

(tail ds)

(tail ds n)

target clj

(target dataset)

Return a dataset containing only the columns that have been marked as inference targets.

train-test-split clj

(train-test-split dataset)

(train-test-split dataset
                  {:keys [train-fraction] :or {train-fraction 0.7} :as options})

Probabilistically split the dataset returning a map of {:train-ds :test-ds}.

Options:

• :randomize-dataset? - When true, shuffle the dataset. In that case 'seed' may be provided. Defaults to true.
• :seed - when :randomize-dataset? is true then this can either be an implementation of java.util.Random or an integer seed which will be used to construct java.util.Random.
• :train-fraction - Fraction of the dataset to use as training set. Defaults to 0.7.

transform-categorical-map clj

(transform-categorical-map dataset fit-data)

Apply a categorical mapping transformation fit with fit-categorical-map.

transform-one-hot clj

(transform-one-hot dataset one-hot-fit-data)

Apply a one-hot transformation to a dataset

ungroup clj

(ungroup ds)

(ungroup ds
         {:keys [order? add-group-as-column add-group-id-as-column separate?
                 dataset-name parallel?]
          :or {separate? true}})

Concat groups into dataset.

When add-group-as-column or add-group-id-as-column is set to true or name(s), columns with group name(s) or group id is added to the result.

Before joining the groups groups can be sorted by group name.

union clj

(union ds & datasets)

unique-by clj

(unique-by ds)

(unique-by ds columns-selector)

(unique-by
  ds
  columns-selector
  {:keys [strategy select-keys parallel?] :or {strategy :first} :as options})

unmark-group clj

(unmark-group ds)

Remove grouping tag

unroll clj

(unroll ds columns-selector)

(unroll ds columns-selector options)

update-columns clj

(update-columns ds columns-map)

(update-columns ds columns-selector update-functions)

write! clj

(write! dataset output-path)

(write! dataset output-path options)

Write a dataset out to a file. Supported forms are:

(ds/write! test-ds "test.csv")
(ds/write! test-ds "test.tsv")
(ds/write! test-ds "test.tsv.gz")
(ds/write! test-ds "test.nippy")
(ds/write! test-ds out-stream)

Options:

• :max-chars-per-column - csv,tsv specific, defaults to 65536 - values longer than this will cause an exception during serialization.
• :max-num-columns - csv,tsv specific, defaults to 8192 - If the dataset has more than this number of columns an exception will be thrown during serialization.
• :quoted-columns - csv specific - sequence of columns names that you would like to always have quoted.
• :file-type - Manually specify the file type. This is usually inferred from the filename but if you pass in an output stream then you will need to specify the file type.
• :headers? - if csv headers are written, defaults to true.

write-csv! clj

write-nippy! clj

(write-nippy! ds filename)