paddle_quantum.data_analysis.vqr

The VQR model.

paddle_quantum.data_analysis.vqr.load_dataset(data_file, model_name)

Loading the Kaggle regression model

Parameters:

  • data_file (str) – Dataset file name.

  • model_name (str) – should be either linear or poly .

Returns:

Raw data.

paddle_quantum.data_analysis.vqr.IPEstimator(circuit, input_state, measure_idx=[0])

Kernel-1 using direct encoded data state to evaluate inner product

Parameters:

  • circuit (Circuit) – Executed quantum circuit

  • input_state (State) – Input state. Defaults to None.

Returns:

The value of inner product

Return type:

Tensor

class paddle_quantum.data_analysis.vqr.QRegressionModel(data_file, model_name, x_feature, y_feature, num_variable, init_params, num_qubits=6, learning_rate=0.1, iteration=100, language='CN')

Bases: object

Regression model covering all classes.

Parameters:

  • data_file (str) – The dataset.csv file.

  • model_name (str) – The regression model.

  • x_feature (List[str]) – Independent variables from data labels.

  • y_feature (str) – Dependent feature values.

  • num_variable (int) – The number of variable initialized in the model.

  • init_params (List[float]) – The initial parameters.

  • num_qubits (int) – The number of qubits in the estimator. Defaults to 6.

  • learning_rate (float) – The learning rate. Defaults to 0.1.

  • iteration (int) – The number optimization steps. Defaults to 100.

  • language (str) – The print language, Defaults to CN .

regression_analyse()
class paddle_quantum.data_analysis.vqr.LinearRegression(num_qubits, num_x=1)

Bases: Layer

Regression class for initializing a quantum linear regression model

Parameters:

  • num_qubits (int) – The number of qubits which the quantum circuit contains.

  • num_x (int) – The the number of independent variables of data. Defaults to 1.

property reg_param: Tensor

Flattened parameters in the Layer.

set_params(new_params)

set parameters of the model.

Parameters:

params – New parameters

fit(X, y, learning_rate=0.01, iteration=200, saved_dir='', model_name='linear')

Fitting method used for training the model

Parameters:

  • X (Tensor | ndarray) – Independent data in a 2D array.

  • y (Tensor | ndarray) – Dependent data in an 1D array.

  • learning_rate (float) – Learning rate of optimization. Defaults to 0.01.

  • iteration (int) – Total training iteration. Defaults to 200.

  • saved_dir (str) – The path for saving the fitted data. Defaults to ''.

  • model_name (str) – The model name. Defaults to linear.

Returns:

Trained model

Return type:

None

predict(X)

Predict value based on current model parameters

Parameters:

X (Tensor | ndarray) – Independent data in a 2D array. Every column indicates an independent variable. Every row indicates a sample of data.

Returns:

predicted value

Return type:

Tensor | ndarray

score(X, y, metric='r2_score')

Quantifying the quality of predictions given test set

Parameters:

  • X (Tensor | ndarray) – Independent data in a 2D array. Every column indicates an independent variable. Every row indicates a sample of data.

  • y (Tensor | ndarray) – Dependent data in a 1D array.

  • metric (str | Callable) – The metric name for the quality. Defaults to r2. If the metric is a callable function, the function should be in the expression

  • function(y_true – np.ndarray, y_pred: np.ndarray) -> float.

Returns:

The model score. Based on sklearn.metric class.

Return type:

float

class paddle_quantum.data_analysis.vqr.PolyRegression(num_qubits, order=1)

Bases: Layer

Regression class for initializing a quantum polynomial regression model

Parameters:

  • num_qubits (int) – The number of qubits which the quantum circuit contains.

  • order (int) – The order of the polynomial regression model. Defaults to 1.

property reg_param: Tensor

Flattened parameters in the Layer.

set_params(new_params)

set parameters of the model.

Parameters:

params – New parameters

fit(X, y, learning_rate=0.01, iteration=200, saved_dir='', model_name='poly')

Fitting method used for training the model

Parameters:

  • X (Tensor | ndarray) – Independent data in a 2D array.

  • y (Tensor | ndarray) – Dependent data in an 1D array.

  • learning_rate (float) – Learning rate of optimization. Defaults to 0.01.

  • iteration (int) – Total training iteration. Defaults to 200.

  • saved_dir (str) – The path for saving the fitted data. Defaults to ''.

  • model_name (str) – The model name. Defaults to poly.

Returns:

Trained model

Return type:

None

predict(X)

Predict value based on current model parameters

Parameters:

x – A sample of data in an array.

Returns:

predicted value

Return type:

Tensor | ndarray

score(X, y, metric='r2_score')

Quantifying the quality of predictions given test set

Parameters:

  • X (Tensor | ndarray) – Independent data in a 2D array. Every column indicates an independent variable. Every row indicates a sample of data.

  • y (Tensor | ndarray) – Dependent data in a 1D array.

  • metric (str | Callable) – The metric name for the quality. Defaults to r2. If the metric is a callable function, the function should be in the expression

  • function(y_true – np.ndarray, y_pred: np.ndarray) -> float.

Returns:

The model score. Based on sklearn.metric class.

Return type:

float