Skip to content

Latest commit

 

History

History
59 lines (49 loc) · 2.43 KB

BuildingData.md

File metadata and controls

59 lines (49 loc) · 2.43 KB
Raw

Building a Dataset

A Wisteria::Data::Dataset object can be built programmatically, as well as imported. To build a dataset, you first construct it as a shared_ptr<Data::Dataset>, then you set its data via AddRow(). After that, if you are using a categorical (e.g., grouping) variable, you can set its string labels.

The following demonstrates this:

auto yData = std::make_shared<Data::Dataset>();

// specify the columns
yData->AddContinuousColumn(L"Score");
yData->AddCategoricalColumn(L"Group");
// fill the continuous column with values
// and set the group ID to 0
yData->AddRow(RowInfo().Continuous({ 13.2 }).Categoricals({ 0 }));
yData->AddRow(RowInfo().Continuous({ 12 }).Categoricals({ 0 }));
yData->AddRow(RowInfo().Continuous({ 11.1 }).Categoricals({ 0 }));
yData->AddRow(RowInfo().Continuous({ 9.9 }).Categoricals({ 0 }));
yData->AddRow(RowInfo().Continuous({ 12 }).Categoricals({ 0 }));
yData->AddRow(RowInfo().Continuous({ 8.75 }).Categoricals({ 0 }));
yData->AddRow(RowInfo().Continuous({ 12 }).Categoricals({ 0 }));

// continue filling the continuous column,
// but set these observations' group IDs to 1
yData->AddRow(RowInfo().Continuous({ 7.1 }).Categoricals({ 1 }));
yData->AddRow(RowInfo().Continuous({ 25 }).Categoricals({ 1 }));
yData->AddRow(RowInfo().Continuous({ 13 }).Categoricals({ 1 }));
yData->AddRow(RowInfo().Continuous({ 10 }).Categoricals({ 1 }));
yData->AddRow(RowInfo().Continuous({ 11 }).Categoricals({ 1 }));
yData->AddRow(RowInfo().Continuous({ 13 }).Categoricals({ 1 }));

// set the string labels for the codes in the grouping column
yData->GetCategoricalColumn(L"Group")->GetStringTable() =
    { { 0, L"MALE" }, { 1, L"FEMALE" } };

More direct access to the dataset is also available, where you can copy raw data into it. The following demonstrates this:

auto yData = std::make_shared<Data::Dataset>();

// specify the columns
yData->AddContinuousColumn(L"Score");
yData->AddCategoricalColumn(L"Group");
// resize the new columns to 10 rows
yData->Resize(10);
// set the categorical codes
yData->GetCategoricalColumn(L"Group")->GetStringTable() =
    { { 0, L"MALE" }, { 1, L"FEMALE" } };
// fill the columns with data
yData->GetCategoricalColumn(L"Group")->Fill({ 1, 1, 1, 2, 1, 2, 2, 1, 2, 1 });
yData->GetContinuousColumn(L"Score")->Fill({ 81, 82, 93, 77, 92, 89, 97, 82, 68, 92 });

After the dataset is built, you can pass it to most graph types to plot it.