Signature | Description | Parameters |
---|---|---|
#include <DataFrame/DataFrameFinancialVisitors.h> template<typename T, typename I = unsigned long, std::size_t A = 0> struct HurstExponentVisitor; // ------------------------------------- template<typename T, typename I = unsigned long, std::size_t A = 0> using hexpo_v = HurstExponentVisitor<T, I, A>; |
This is a “single action visitor”, meaning it is passed the whole data vector in one call and you must use the single_act_visit() interface. This functor calculates the Hurst exponent for the given column. We can use the Hurst exponent (H) as a measure for long-term memory of a time series, that is, to measure the amount by which that series deviates from a random walk. The scalar represents the relative tendency of a time series either to regress strongly to the mean (mean-reverting pattern) or to cluster in a certain direction (trending pattern). The values of the Hurst exponent range between 0 and 1. Based on the value of H, we can classify any time series into one of the three categories:
using RangeVec = std::vector<size_type, typename allocator_declare<size_type, A>::type>; explicit HurstExponentVisitor(RangeVec<size_t> &&ranges); ranges: A vector of column length divisors. For example, {1, 2, 4 } means calculate Hurst exponent in 3 steps. It divides the time-series column to 1 chunk, 2 chunks and 4 chunks. |
T: Column data type. I: Index type. A: Memory alignment boundary for vectors. Default is system default alignment |
static void test_HurstExponentVisitor() { std::cout << "\nTesting HurstExponentVisitor{ } ..." << std::endl; RandGenParams<double> p; p.seed = 123; p.min_value = 0; p.max_value = 30; std::vector<double> d1 = gen_uniform_real_dist<double>(1024, p); std::vector<double> d2 = { 0.04, 0.02, 0.05, 0.08, 0.02, -0.17, 0.05, 0.0 }; std::vector<double> d3 = { 0.04, 0.05, 0.055, 0.06, 0.061, 0.072, 0.073, 0.8 }; MyDataFrame df; df.load_index(std::move(MyDataFrame::gen_sequence_index(0, 1024, 1))); df.load_column("d1_col", std::move(d1), nan_policy::dont_pad_with_nans); df.load_column("d2_col", std::move(d2), nan_policy::dont_pad_with_nans); df.load_column("d3_col", std::move(d3), nan_policy::dont_pad_with_nans); HurstExponentVisitor<double> he_v1 ({ 1, 2, 4 }); auto result1 = df.single_act_visit<double>("d2_col", he_v1).get_result(); assert(result1 - 0.865926 < 0.00001); HurstExponentVisitor<double> he_v2 ({ 1, 2, 4, 5, 6, 7 }); auto result2 = df.single_act_visit<double>("d1_col", he_v2).get_result(); assert(result2 - 0.487977 < 0.00001); HurstExponentVisitor<double> he_v3 ({ 1, 2, 4 }); auto result3 = df.single_act_visit<double>("d3_col", he_v3).get_result(); assert(result3 - 0.903057 < 0.00001); }