SimonBlanke /
Hyperactive
| @@ 10-88 (lines=79) @@ | ||
| 7 | import pandas as pd |
|
| 8 | ||
| 9 | ||
| 10 | class Results: |
|
| 11 | def __init__(self, results_list, opt_pros): |
|
| 12 | self.results_list = results_list |
|
| 13 | self.opt_pros = opt_pros |
|
| 14 | ||
| 15 | self.objFunc2results = {} |
|
| 16 | self.search_id2results = {} |
|
| 17 | ||
| 18 | def _sort_results_objFunc(self, objective_function): |
|
| 19 | best_score = -np.inf |
|
| 20 | best_para = None |
|
| 21 | search_data = None |
|
| 22 | ||
| 23 | search_data_list = [] |
|
| 24 | ||
| 25 | for results_ in self.results_list: |
|
| 26 | nth_process = results_["nth_process"] |
|
| 27 | ||
| 28 | opt = self.opt_pros[nth_process] |
|
| 29 | objective_function_ = opt.experiment.objective_function |
|
| 30 | search_space_ = opt.s_space() |
|
| 31 | params = list(search_space_.keys()) |
|
| 32 | ||
| 33 | if objective_function_ != objective_function: |
|
| 34 | continue |
|
| 35 | ||
| 36 | if results_["best_score"] > best_score: |
|
| 37 | best_score = results_["best_score"] |
|
| 38 | best_para = results_["best_para"] |
|
| 39 | ||
| 40 | search_data = results_["search_data"] |
|
| 41 | search_data["eval_times"] = results_["eval_times"] |
|
| 42 | search_data["iter_times"] = results_["iter_times"] |
|
| 43 | ||
| 44 | search_data_list.append(search_data) |
|
| 45 | ||
| 46 | if len(search_data_list) > 0: |
|
| 47 | search_data = pd.concat(search_data_list) |
|
| 48 | ||
| 49 | self.objFunc2results[objective_function] = { |
|
| 50 | "best_para": best_para, |
|
| 51 | "best_score": best_score, |
|
| 52 | "search_data": search_data, |
|
| 53 | "params": params, |
|
| 54 | } |
|
| 55 | ||
| 56 | def _get_result(self, id_, result_name): |
|
| 57 | if id_ not in self.objFunc2results: |
|
| 58 | self._sort_results_objFunc(id_) |
|
| 59 | ||
| 60 | search_data = self.objFunc2results[id_][result_name] |
|
| 61 | ||
| 62 | return search_data |
|
| 63 | ||
| 64 | def best_para(self, id_): |
|
| 65 | best_para_ = self._get_result(id_, "best_para") |
|
| 66 | ||
| 67 | if best_para_ is not None: |
|
| 68 | return best_para_ |
|
| 69 | ||
| 70 | raise ValueError("objective function name not recognized") |
|
| 71 | ||
| 72 | def best_score(self, id_): |
|
| 73 | best_score_ = self._get_result(id_, "best_score") |
|
| 74 | ||
| 75 | if best_score_ != -np.inf: |
|
| 76 | return best_score_ |
|
| 77 | ||
| 78 | raise ValueError("objective function name not recognized") |
|
| 79 | ||
| 80 | def search_data(self, id_): |
|
| 81 | search_data = self._get_result(id_, "search_data") |
|
| 82 | ||
| 83 | params = self.objFunc2results[id_]["params"] |
|
| 84 | ||
| 85 | if search_data is not None: |
|
| 86 | return search_data |
|
| 87 | ||
| 88 | raise ValueError("objective function name not recognized") |
|
| 89 | ||
| @@ 10-88 (lines=79) @@ | ||
| 7 | import pandas as pd |
|
| 8 | ||
| 9 | ||
| 10 | class Results: |
|
| 11 | def __init__(self, results_list, opt_pros): |
|
| 12 | self.results_list = results_list |
|
| 13 | self.opt_pros = opt_pros |
|
| 14 | ||
| 15 | self.objFunc2results = {} |
|
| 16 | self.search_id2results = {} |
|
| 17 | ||
| 18 | def _sort_results_objFunc(self, objective_function): |
|
| 19 | best_score = -np.inf |
|
| 20 | best_para = None |
|
| 21 | search_data = None |
|
| 22 | ||
| 23 | search_data_list = [] |
|
| 24 | ||
| 25 | for results_ in self.results_list: |
|
| 26 | nth_process = results_["nth_process"] |
|
| 27 | ||
| 28 | opt = self.opt_pros[nth_process] |
|
| 29 | objective_function_ = opt.objective_function |
|
| 30 | search_space_ = opt.s_space() |
|
| 31 | params = list(search_space_.keys()) |
|
| 32 | ||
| 33 | if objective_function_ != objective_function: |
|
| 34 | continue |
|
| 35 | ||
| 36 | if results_["best_score"] > best_score: |
|
| 37 | best_score = results_["best_score"] |
|
| 38 | best_para = results_["best_para"] |
|
| 39 | ||
| 40 | search_data = results_["search_data"] |
|
| 41 | search_data["eval_times"] = results_["eval_times"] |
|
| 42 | search_data["iter_times"] = results_["iter_times"] |
|
| 43 | ||
| 44 | search_data_list.append(search_data) |
|
| 45 | ||
| 46 | if len(search_data_list) > 0: |
|
| 47 | search_data = pd.concat(search_data_list) |
|
| 48 | ||
| 49 | self.objFunc2results[objective_function] = { |
|
| 50 | "best_para": best_para, |
|
| 51 | "best_score": best_score, |
|
| 52 | "search_data": search_data, |
|
| 53 | "params": params, |
|
| 54 | } |
|
| 55 | ||
| 56 | def _get_result(self, id_, result_name): |
|
| 57 | if id_ not in self.objFunc2results: |
|
| 58 | self._sort_results_objFunc(id_) |
|
| 59 | ||
| 60 | search_data = self.objFunc2results[id_][result_name] |
|
| 61 | ||
| 62 | return search_data |
|
| 63 | ||
| 64 | def best_para(self, id_): |
|
| 65 | best_para_ = self._get_result(id_, "best_para") |
|
| 66 | ||
| 67 | if best_para_ is not None: |
|
| 68 | return best_para_ |
|
| 69 | ||
| 70 | raise ValueError("objective function name not recognized") |
|
| 71 | ||
| 72 | def best_score(self, id_): |
|
| 73 | best_score_ = self._get_result(id_, "best_score") |
|
| 74 | ||
| 75 | if best_score_ != -np.inf: |
|
| 76 | return best_score_ |
|
| 77 | ||
| 78 | raise ValueError("objective function name not recognized") |
|
| 79 | ||
| 80 | def search_data(self, id_): |
|
| 81 | search_data = self._get_result(id_, "search_data") |
|
| 82 | ||
| 83 | params = self.objFunc2results[id_]["params"] |
|
| 84 | ||
| 85 | if search_data is not None: |
|
| 86 | return search_data |
|
| 87 | ||
| 88 | raise ValueError("objective function name not recognized") |
|
| 89 | ||
