| 45 | } |
| 46 | |
| 47 | void Popularity::run(obj_info_map_type &obj_map) { |
| 48 | /* freq_vec_ is a sorted vec of obj frequency */ |
| 49 | freq_vec_.reserve(obj_map.size()); |
| 50 | for (const auto &p : obj_map) { |
| 51 | freq_vec_.push_back(p.second.freq); |
| 52 | } |
| 53 | sort(freq_vec_.begin(), freq_vec_.end(), greater<>()); |
| 54 | |
| 55 | if (obj_map.size() < 200) { |
| 56 | fit_fail_reason_ = "popularity: too few objects (" + |
| 57 | to_string(obj_map.size()) + |
| 58 | "), skip the popularity computation"; |
| 59 | WARN("%s\n", fit_fail_reason_.c_str()); |
| 60 | return; |
| 61 | } |
| 62 | |
| 63 | if (freq_vec_[0] < 200) { |
| 64 | fit_fail_reason_ = "popularity: the most popular object has " + |
| 65 | to_string(freq_vec_[0]) + " requests "; |
| 66 | WARN("%s\n", fit_fail_reason_.c_str()); |
| 67 | } |
| 68 | |
| 69 | /* calculate Zipf alpha using linear regression */ |
| 70 | vector<double> log_freq(obj_map.size()); |
| 71 | vector<double> log_rank(obj_map.size()); |
| 72 | |
| 73 | int i = 0; |
| 74 | for_each(log_freq.begin(), log_freq.end(), |
| 75 | [&](double &item) { item = log(freq_vec_[i++]); }); |
| 76 | i = 0; |
| 77 | for_each(log_rank.begin(), log_rank.end(), [&](double &item) { |
| 78 | ++i; |
| 79 | item = log(i); |
| 80 | }); |
| 81 | |
| 82 | /* TODO: a better linear regression with intercept and R2 */ |
| 83 | slope_ = -PopularityUtils::slope(log_rank, log_freq); |
| 84 | |
| 85 | has_run = true; |
| 86 | } |
| 87 | |
| 88 | vector<uint32_t> freq_vec_{}; |
| 89 | double slope_, intercept_, r2_; |