Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Added LGBM_DatasetCreateFromFile #164

Open
wants to merge 10 commits into
base: all-params
Choose a base branch
from
Open

Added LGBM_DatasetCreateFromFile #164

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
753fc94
implemented LGBM_DatasetCreateFromFile
kainkad Oct 18, 2024
155de38
some unclear behaviour
kainkad Oct 18, 2024
c2340fb
further curiosities
kainkad Oct 22, 2024
617a788
updated comments
kainkad Oct 28, 2024
aa04d9e
:pencil2:
kainkad Oct 28, 2024
83a17ce
updated fitting method to include datasets from filepaths
kainkad Nov 5, 2024
02613ff
:pencil2: bye bye
kainkad Nov 7, 2024
8d5f3ee
updated with the target branch
kainkad Nov 11, 2024
d31e144
updated with all-params
kainkad Nov 26, 2024
2712934
updated params
kainkad Nov 27, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 17 additions & 1 deletion src/MLJInterface.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,6 +70,7 @@ MLJModelInterface.@mlj_model mutable struct LGBMRegressor <: MLJModelInterface.D
max_cat_threshold::Int = 32::(_ > 0)
cat_l2::Float64 = 10.0::(_ >= 0)
cat_smooth::Float64 = 10.0::(_ >= 0)
forcedsplits_filename::String = ""
refit_decay_rate::Float64 = 0.9::(0.0 <= _ <= 1.0)

# Dataset parameters
Expand All @@ -80,7 +81,14 @@ MLJModelInterface.@mlj_model mutable struct LGBMRegressor <: MLJModelInterface.D
is_enable_sparse::Bool = true
use_missing::Bool = true
feature_pre_filter::Bool = true
two_round::Bool = false
header::Bool = false
label_column::String = ""
weight_column::String = ""
ignore_column::String = ""
categorical_feature::Vector{Int} = Vector{Int}()
forcedbins_filename::String = ""
precise_float_parser::Bool = false

# Predict parameters
start_iteration_predict::Int = 0
Expand Down Expand Up @@ -166,6 +174,7 @@ MLJModelInterface.@mlj_model mutable struct LGBMClassifier <: MLJModelInterface.
max_cat_threshold::Int = 32::(_ > 0)
cat_l2::Float64 = 10.0::(_ >= 0)
cat_smooth::Float64 = 10.0::(_ >= 0)
forcedsplits_filename::String = ""
refit_decay_rate::Float64 = 0.9::(0.0 <= _ <= 1.0)

# Dateset parameters
Expand All @@ -176,7 +185,14 @@ MLJModelInterface.@mlj_model mutable struct LGBMClassifier <: MLJModelInterface.
is_enable_sparse::Bool = true
use_missing::Bool = true
feature_pre_filter::Bool = true
categorical_feature::Vector{Int} = Vector{Int}();
two_round::Bool = false
header::Bool = false
label_column::String = ""
weight_column::String = ""
ignore_column::String = ""
categorical_feature::Vector{Int} = Vector{Int}()
forcedbins_filename::String = ""
precise_float_parser::Bool = false

# Predict parameters
start_iteration_predict::Int = 0
Expand Down
90 changes: 81 additions & 9 deletions src/estimators.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,6 +47,7 @@ mutable struct LGBMRegression <: LGBMEstimator
max_cat_threshold::Int
cat_l2::Float64
cat_smooth::Float64
forcedsplits_filename::String
refit_decay_rate::Float64

# Dataset parameters
Expand All @@ -57,7 +58,14 @@ mutable struct LGBMRegression <: LGBMEstimator
is_enable_sparse::Bool
use_missing::Bool
feature_pre_filter::Bool
two_round::Bool
header::Bool
label_column::String
weight_column::String
ignore_column::String
categorical_feature::Vector{Int}
forcedbins_filename::String
precise_float_parser::Bool

# Predict parameters
start_iteration_predict::Int
Expand Down Expand Up @@ -133,6 +141,7 @@ end
max_cat_threshold = 32,
cat_l2 = 10.,
cat_smooth = 10.,
forcedsplits_filename = "",
refit_decay_rate = 0.9,
linear_tree = false,
max_bin = 255,
Expand All @@ -141,7 +150,14 @@ end
is_enable_sparse = true,
use_missing = true,
feature_pre_filter = true,
two_round = false,
header = false,
label_column = "",
weight_column = "",
ignore_column = "",
categorical_feature = Int[],
forcedbins_filename = "",
precise_float_parser = false,
start_iteration_predict = 0,
num_iteration_predict = -1,
predict_raw_score = false,
Expand Down Expand Up @@ -207,6 +223,7 @@ function LGBMRegression(;
max_cat_threshold = 32,
cat_l2 = 10.,
cat_smooth = 10.,
forcedsplits_filename = "",
refit_decay_rate = 0.9,
linear_tree = false,
max_bin = 255,
Expand All @@ -215,7 +232,14 @@ function LGBMRegression(;
is_enable_sparse = true,
use_missing = true,
feature_pre_filter = true,
two_round = false,
header = false,
label_column = "",
weight_column = "",
ignore_column = "",
categorical_feature = Int[],
forcedbins_filename = "",
precise_float_parser = false,
start_iteration_predict = 0,
num_iteration_predict = -1,
predict_raw_score = false,
Expand Down Expand Up @@ -247,9 +271,9 @@ function LGBMRegression(;
extra_seed, early_stopping_round, max_delta_step, lambda_l1, lambda_l2,
min_gain_to_split, drop_rate, max_drop, skip_drop,
xgboost_dart_mode, uniform_drop, drop_seed, top_rate, other_rate, min_data_per_group, max_cat_threshold,
cat_l2, cat_smooth, refit_decay_rate, linear_tree, max_bin, bin_construct_sample_cnt, data_random_seed,
is_enable_sparse, use_missing, feature_pre_filter, categorical_feature,
start_iteration_predict, num_iteration_predict, predict_raw_score, predict_leaf_index, predict_contrib, predict_disable_shape_check,
cat_l2, cat_smooth, forcedsplits_filename, refit_decay_rate, linear_tree, max_bin, bin_construct_sample_cnt, data_random_seed,
is_enable_sparse, use_missing, feature_pre_filter, two_round, header, label_column, weight_column, ignore_column, categorical_feature, forcedbins_filename,
precise_float_parser, start_iteration_predict, num_iteration_predict, predict_raw_score, predict_leaf_index, predict_contrib, predict_disable_shape_check,
1, is_unbalance, boost_from_average, alpha, metric, metric_freq, is_provide_training_metric, eval_at, num_machines, local_listen_port, time_out,
machine_list_filename, gpu_platform_id, gpu_device_id, gpu_use_dp, num_gpu,
)
Expand Down Expand Up @@ -304,6 +328,7 @@ mutable struct LGBMClassification <: LGBMEstimator
max_cat_threshold::Int
cat_l2::Float64
cat_smooth::Float64
forcedsplits_filename::String
refit_decay_rate::Float64

# Dataset parameters
Expand All @@ -314,7 +339,14 @@ mutable struct LGBMClassification <: LGBMEstimator
is_enable_sparse::Bool
use_missing::Bool
feature_pre_filter::Bool
two_round::Bool
header::Bool
label_column::String
weight_column::String
ignore_column::String
categorical_feature::Vector{Int}
forcedbins_filename::String
precise_float_parser::Bool

# Predict parameters
start_iteration_predict::Int
Expand Down Expand Up @@ -396,6 +428,7 @@ end
max_cat_threshold = 32,
cat_l2 = 10.,
cat_smooth = 10.,
forcedsplits_filename = "",
refit_decay_rate = 0.9,
linear_tree = false,
max_bin = 255,
Expand All @@ -404,7 +437,14 @@ end
is_enable_sparse = true,
use_missing = true,
feature_pre_filter = true,
two_round = false,
header = false,
label_column = "",
weight_column = "",
ignore_column = "",
categorical_feature = Int[],
forcedbins_filename = "",
precise_float_parser = false,
start_iteration_predict = 0,
num_iteration_predict = -1,
predict_raw_score = false,
Expand Down Expand Up @@ -477,6 +517,7 @@ function LGBMClassification(;
max_cat_threshold = 32,
cat_l2 = 10.,
cat_smooth = 10.,
forcedsplits_filename = "",
refit_decay_rate = 0.9,
linear_tree = false,
max_bin = 255,
Expand All @@ -485,7 +526,14 @@ function LGBMClassification(;
is_enable_sparse = true,
use_missing = true,
feature_pre_filter = true,
two_round = false,
header = false,
label_column = "",
weight_column = "",
ignore_column = "",
categorical_feature = Int[],
forcedbins_filename = "",
precise_float_parser = false,
start_iteration_predict = 0,
num_iteration_predict = -1,
predict_raw_score = false,
Expand Down Expand Up @@ -521,9 +569,9 @@ function LGBMClassification(;
bagging_fraction, pos_bagging_fraction, neg_bagging_fraction,bagging_freq,
bagging_seed, feature_fraction, feature_fraction_bynode, feature_fraction_seed, extra_trees, extra_seed, early_stopping_round, max_delta_step, lambda_l1, lambda_l2,
min_gain_to_split, drop_rate, max_drop, skip_drop, xgboost_dart_mode,
uniform_drop, drop_seed, top_rate, other_rate, min_data_per_group, max_cat_threshold, cat_l2, cat_smooth, refit_decay_rate, linear_tree, max_bin, bin_construct_sample_cnt,
data_random_seed, is_enable_sparse, use_missing, feature_pre_filter, categorical_feature,
start_iteration_predict, num_iteration_predict, predict_raw_score, predict_leaf_index, predict_contrib,
uniform_drop, drop_seed, top_rate, other_rate, min_data_per_group, max_cat_threshold, cat_l2, cat_smooth, forcedsplits_filename, refit_decay_rate, linear_tree, max_bin, bin_construct_sample_cnt,
data_random_seed, is_enable_sparse, use_missing, feature_pre_filter, two_round, header, label_column, weight_column, ignore_column, categorical_feature, forcedbins_filename,
precise_float_parser, start_iteration_predict, num_iteration_predict, predict_raw_score, predict_leaf_index, predict_contrib,
predict_disable_shape_check, pred_early_stop, pred_early_stop_freq, pred_early_stop_margin,
num_class, is_unbalance, scale_pos_weight, sigmoid, boost_from_average,
metric, metric_freq, is_provide_training_metric, eval_at, num_machines, local_listen_port, time_out,
Expand Down Expand Up @@ -579,6 +627,7 @@ mutable struct LGBMRanking <: LGBMEstimator
max_cat_threshold::Int
cat_l2::Float64
cat_smooth::Float64
forcedsplits_filename::String
refit_decay_rate::Float64

# Dataset parameters
Expand All @@ -589,8 +638,15 @@ mutable struct LGBMRanking <: LGBMEstimator
is_enable_sparse::Bool
use_missing::Bool
feature_pre_filter::Bool
two_round::Bool
header::Bool
label_column::String
weight_column::String
ignore_column::String
group_column::String
categorical_feature::Vector{Int}
forcedbins_filename::String
precise_float_parser::Bool

# Predict parameters
start_iteration_predict::Int
Expand Down Expand Up @@ -676,6 +732,7 @@ end
max_cat_threshold = 32,
cat_l2 = 10.,
cat_smooth = 10.,
forcedsplits_filename = "",
refit_decay_rate = 0.9,
linear_tree = false,
max_bin = 255,
Expand All @@ -684,8 +741,15 @@ end
is_enable_sparse = true,
use_missing = true,
feature_pre_filter = true,
two_round = false,
header = false,
label_column = "",
weight_column = "",
ignore_column = "",
group_column = ""
categorical_feature = Int[],
forcedbins_filename = "",
precise_float_parser = false,
start_iteration_predict = 0,
num_iteration_predict = -1,
predict_raw_score = false,
Expand Down Expand Up @@ -762,6 +826,7 @@ function LGBMRanking(;
max_cat_threshold = 32,
cat_l2 = 10.,
cat_smooth = 10.,
forcedsplits_filename = "",
refit_decay_rate = 0.9,
linear_tree = false,
max_bin = 255,
Expand All @@ -770,8 +835,15 @@ function LGBMRanking(;
is_enable_sparse = true,
use_missing = true,
feature_pre_filter = true,
two_round = false,
header = false,
label_column = "",
weight_column = "",
ignore_column = "",
group_column = "",
categorical_feature = Int[],
forcedbins_filename = "",
precise_float_parser = false,
start_iteration_predict = 0,
num_iteration_predict = -1,
predict_raw_score = false,
Expand Down Expand Up @@ -811,9 +883,9 @@ function LGBMRanking(;
bagging_fraction, pos_bagging_fraction, neg_bagging_fraction, bagging_freq,
bagging_seed, feature_fraction, feature_fraction_bynode, feature_fraction_seed, extra_trees, extra_seed, early_stopping_round, max_delta_step, lambda_l1, lambda_l2,
min_gain_to_split, drop_rate, max_drop, skip_drop, xgboost_dart_mode,
uniform_drop, drop_seed, top_rate, other_rate, min_data_per_group, max_cat_threshold, cat_l2, cat_smooth, refit_decay_rate, linear_tree, max_bin, bin_construct_sample_cnt,
data_random_seed, is_enable_sparse, use_missing, feature_pre_filter, group_column, categorical_feature,
start_iteration_predict, num_iteration_predict, predict_raw_score, predict_leaf_index, predict_contrib,
uniform_drop, drop_seed, top_rate, other_rate, min_data_per_group, max_cat_threshold, cat_l2, cat_smooth, forcedsplits_filename, refit_decay_rate, linear_tree, max_bin, bin_construct_sample_cnt,
data_random_seed, is_enable_sparse, use_missing, feature_pre_filter, two_round, header, label_column, weight_column, ignore_column, group_column, categorical_feature, forcedbins_filename,
precise_float_parser, start_iteration_predict, num_iteration_predict, predict_raw_score, predict_leaf_index, predict_contrib,
predict_disable_shape_check, pred_early_stop, pred_early_stop_freq, pred_early_stop_margin,
objective_seed, num_class, is_unbalance, scale_pos_weight, sigmoid, boost_from_average, lambdarank_truncation_level, lambdarank_norm, label_gain,
metric, metric_freq, is_provide_training_metric, eval_at, num_machines, local_listen_port, time_out,
Expand Down
16 changes: 15 additions & 1 deletion src/wrapper.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -131,7 +131,21 @@ macro lightgbm(f, params...)
end


# function LGBM_DatasetCreateFromFile()
function LGBM_DatasetCreateFromFile(
filename::String,
parameters::String,
reference::Dataset = Dataset(C_NULL)
)
out = Ref{DatasetHandle}()
@lightgbm(
:LGBM_DatasetCreateFromFile,
filename => Cstring,
parameters => Cstring,
reference.handle => DatasetHandle,
out => Ref{DatasetHandle}
)
return Dataset(out[])
end


function LGBM_DatasetCreateFromCSC(
Expand Down
41 changes: 41 additions & 0 deletions test/ffi/datasets.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -74,6 +74,47 @@ end
end


@testset "LGBM_DatasetCreateFromFile" begin
# Create a sample .csv file with a header row
sample_data = """
feature1,feature2,feature3,feature4,feature5
0.1,0.2,0.3,0.4,0.5
0.6,0.7,0.8,0.9,1.0
1.1,1.2,1.3,1.4,1.5
"""
sample_file = "sample_data.csv"
open(sample_file, "w") do f
write(f, sample_data)
end

# Define dataset parameters as strings
Copy link
Contributor Author

@kainkad kainkad Oct 18, 2024 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There's some unclear behaviour for the ignore_column parameter. According to the docs it can be specified for the text files just like the header (which works) and the other params in this test: https://lightgbm.readthedocs.io/en/v3.3.5/Parameters.html#ignore_column. Despite mentioning "text files", only CSV, TSV, and LibSVM (zero-based) are supported (https://github.com/microsoft/LightGBM/blob/master/src/io/parser.cpp#L264-L266) and the sample_data seems to be a valid example and other params do work with it.

params = [
"two_round=true header=true label_column=0 weight_column=1 ignore_column=2",
"two_round=false header=true label_column=1 weight_column=2 ignore_column=3"
]

expected_num_data = [3, 3]
expected_num_feature = [4, 4]
kainkad marked this conversation as resolved.
Show resolved Hide resolved

for (i, param) in enumerate(params)
# Create dataset from file
dataset = LightGBM.LGBM_DatasetCreateFromFile(sample_file, param)

# Check if dataset is created successfully
@test dataset != C_NULL

# Check the number of rows and columns
num_data = LightGBM.LGBM_DatasetGetNumData(dataset)
num_feature = LightGBM.LGBM_DatasetGetNumFeature(dataset)
@test num_data == expected_num_data[i]
@test num_feature == expected_num_feature[i]
end

# Clean up
rm(sample_file)
end


@testset "LGBM_DatasetCreateFromCSC" begin

mymat = sparse([1. 2.; 3. 4.; 5. 6.])
Expand Down
Loading