[ENH] Catch22 experimentation (#257)

* wip dir

* [Catch22] Fixing Issue #1384 (#264)

* Initial Testting

* Created CSV and XLSX files with error indication

* fixes

* string fix

* initial changes

* Final changes of today, CO_f1ecac still not giving correct results

* small fix

* Completely Fixed _FC_Local_Simple_mean3_stderr, however _co_f1.. has an autocorrelation error again

* Fixed sb_mean and sb_diff

* Added comment to explain long_stretch

* Renaming of variables for SB_binary

* Completely fixed SB_MotifThree_quantile_hh

* Fixed _CO_f1ecac

* Fixed IN_AutoMutualInfoStats_40_gaussian_fmmi

* Completely fixed Co_embed2

* Fixed all methods when using training data (Final Catch22: SB_transitionMatrix fixed)

* Changed ordering to be the same as using pycatch22, fixed some other small code writing

* Code quality

* fixed catch22 code quality

---------

Co-authored-by: Daniele Carli <47296443+Moonzyyy@users.noreply.github.com>

tsml_eval/_wip/catch22/_test_catch22.py

@@ -0,0 +1,157 @@
+from _catch22 import Catch22
+import pycatch22
+import numpy as np
+from aeon.datasets import tsc_datasets, load_italy_power_demand
+from openpyxl import Workbook
+from openpyxl.styles import PatternFill
+from decimal import Decimal, ROUND_HALF_UP
+import csv
+import os
+
+features_test =  [
+        "mode_5",
+        "mode_10",
+        "stretch_high",
+        "outlier_timing_pos",
+        "outlier_timing_neg",
+        "acf_timescale",
+        "acf_first_min",
+        "centroid_freq",
+        "low_freq_power",
+        "forecast_error",
+        "trev",
+        "ami2",
+        "ami_timescale",
+        "high_fluctuation",
+        "stretch_decreasing",
+        "entropy_pairs",
+        "whiten_timescale",
+        "periodicity",
+        "dfa",
+        "rs_range",
+        "transition_matrix",
+        "periodicity",
+    ]
+
+feature_names_aeon = [
+    "DN_HistogramMode_5",
+    "DN_HistogramMode_10",
+    "SB_BinaryStats_diff_longstretch0",
+    "DN_OutlierInclude_p_001_mdrmd",
+    "DN_OutlierInclude_n_001_mdrmd",
+    "CO_f1ecac",
+    "CO_FirstMin_ac",
+    "SP_Summaries_welch_rect_area_5_1",
+    "SP_Summaries_welch_rect_centroid",
+    "FC_LocalSimple_mean3_stderr",
+    "CO_trev_1_num",
+    "CO_HistogramAMI_even_2_5",
+    "IN_AutoMutualInfoStats_40_gaussian_fmmi",
+    "MD_hrv_classic_pnn40",
+    "SB_BinaryStats_mean_longstretch1",
+    "SB_MotifThree_quantile_hh",
+    "FC_LocalSimple_mean1_tauresrat",
+    "CO_Embed2_Dist_tau_d_expfit_meandiff",
+    "SC_FluctAnal_2_dfa_50_1_2_logi_prop_r1",
+    "SC_FluctAnal_2_rsrangefit_50_1_logi_prop_r1",
+    "SB_TransitionMatrix_3ac_sumdiagcov",
+    "PD_PeriodicityWang_th0_01",
+]
+
+features_names_pycatch22 = [
+        'DN_HistogramMode_5',
+        'DN_HistogramMode_10',
+        'CO_f1ecac',
+        'CO_FirstMin_ac',
+        'CO_HistogramAMI_even_2_5',
+        'CO_trev_1_num',
+        'MD_hrv_classic_pnn40',
+        'SB_BinaryStats_mean_longstretch1',
+        'SB_TransitionMatrix_3ac_sumdiagcov',
+        'PD_PeriodicityWang_th0_01',
+        'CO_Embed2_Dist_tau_d_expfit_meandiff',
+        'IN_AutoMutualInfoStats_40_gaussian_fmmi',
+        'FC_LocalSimple_mean1_tauresrat',
+        'DN_OutlierInclude_p_001_mdrmd',
+        'DN_OutlierInclude_n_001_mdrmd',
+        'SP_Summaries_welch_rect_area_5_1',
+        'SB_BinaryStats_diff_longstretch0',
+        'SB_MotifThree_quantile_hh',
+        'SC_FluctAnal_2_rsrangefit_50_1_logi_prop_r1',
+        'SC_FluctAnal_2_dfa_50_1_2_logi_prop_r1',
+        'SP_Summaries_welch_rect_centroid',
+        'FC_LocalSimple_mean3_stderr'
+    ]
+
+# $env:NUMBA_DISABLE_JIT = "0"
+# echo $env:NUMBA_DISABLE_JIT
+# Numba Disabled Switcher 0 = off, 1 = on
+
+IPD_X_train, IPD_y_train = load_italy_power_demand(split="train")
+#IPD_X_train = [IPD_X_train[332]]
+#print("Training Data: ", IPD_X_train)
+os.environ['NUMBA_DISABLE_JIT'] = '0'
+print("Numba Off: ",os.environ['NUMBA_DISABLE_JIT'])
+
+aeon_file_name = ""
+if os.environ['NUMBA_DISABLE_JIT'] == '0':
+    aeon_file_name = "aeon_catch22_with_numba"
+else:
+    aeon_file_name = "aeon_catch22_no_numba"
+#aeon
+aeon_c22 = Catch22(features=features_test,replace_nans=True)
+_ = aeon_c22.fit_transform(IPD_X_train)
+
+results_aeon = [features_test]
+for i in range(len(_)):
+    #formatting it to pycatch22 format
+    results_aeon.append(_[i])
+
+
+#pycatch
+results_pycatch22 = [features_names_pycatch22]
+for i in range(len(IPD_X_train)):
+    results = pycatch22.catch22_all(IPD_X_train[i][0])
+    results_pycatch22.append(results['values'])
+
+
+#aeon xlsx
+wb = Workbook()
+ws = wb.active
+for i in range(len(results_aeon)):
+    for j in range(len(results_aeon[i])):
+        cell = ws.cell(row=i + 1, column=j + 1, value=results_aeon[i][j])
+        if(i != 0):
+            rounded_aeon_results = Decimal(results_aeon[i][j]).quantize(Decimal('.0000'), 
+                                                                    rounding=ROUND_HALF_UP)
+            rounded_pycatch22_results = Decimal(results_pycatch22[i][j]).quantize(Decimal('.0000'), 
+                                                                                rounding=ROUND_HALF_UP)
+            if rounded_aeon_results != rounded_pycatch22_results:
+                cell.fill = PatternFill(start_color="FF0000", end_color="FF0000", fill_type="solid")
+
+
+wb.save(aeon_file_name + ".xlsx")
+
+
+#pycatch xlsx
+wb = Workbook()
+ws = wb.active
+for i in range(len(results_pycatch22)):
+    for j in range(len(results_pycatch22[i])):
+        cell = ws.cell(row=i + 1, column=j + 1, value=results_pycatch22[i][j])
+wb.save("pycatch22_catch22_ipd.xlsx")
+
+
+#aeon.csv
+with open(aeon_file_name + ".csv", mode='w', newline='') as file:
+    writer = csv.writer(file)
+    for data in results_aeon:
+        writer.writerow(data)
+
+#pycatch.csv  
+with open('pycatch22_catch22_ipd.csv', mode='w', newline='') as file:
+    writer = csv.writer(file)
+    for data in results_pycatch22:
+        writer.writerow(data)
+
+print("Finished writing data")