![[PukiWiki]](image/pukiwiki.png "[PukiWiki]")
↓質問者チェッカー
議事録テンプレ↓
new[i]}':f'{time_new[i]+datetime.timedelta(minutes=15)}']))\n C_score.append(math.log(size[i]))\n\n C_score_max = max(C_score, default = C_score)\n #print(C_score)\n\n #C_score_norm = []\n # for j in range(len(time_new)):\n # for i in range(len(df_i)):\n # if time_new[j] < df_i.index[i] < time_new[j]+datetime.timedelta(minutes=15):\n # C_score_norm.append(5 * C_score[j] / C_score_max)\n # else:\n # C_score_norm.append(0)\n\n #print(len(time_new))\n if len(time_new) >= 1:\n for j in range(len(time_new)):\n for i in range(len(df_i)):\n if time_new[j] < df_i.index[i] < time_new[j]+datetime.timedelta(minutes=15):\n #C_score_norm.append(5 * C_score[j] / C_score_max)\n C_score_norm[i] = (5 * C_score[j] / C_score_max)\n else:\n #C_score_norm.append(0)\n C_score_norm[i] = 0\n else:\n pass\n\n return C_score_norm\n\n###########################\n###########################\n\n###video版の集中性スコア#####\ndef Video_Concentration(df):\n\n df_a = df.loc[:,['evaluation','date']]\n\n for i in range(len(df_a.index)):\n #print(df_a.iat[i,1])\n df_a.iat[i,1] = datetime.datetime.strptime(df_a.iat[i,1], '%Y-%m-%d %H:%M:%S')\n\n df_i = df_a.set_index('date')\n\n #aaa = df_i[f'{date} 00:00:00':f'{date} 23:59:59']\n #print(aaa)\n\n df_s = df_i.sort_index()\n\n #evaluation が５のものを取り出す\n df_aaa = df_s.query('evaluation == 5')\n\n pd.set_option('display.max_rows', 10000)\n\n #日付ごとに評価値５のものの個数を出力する\n #例：2022-02-14では評価値５のレビューが5回あった， 2022-02-15では評価値５のレビューが1回\n #それを，データフレームに入れる → NANの部分が全て０になる\n df_n = pd.DataFrame(df_aaa.resample('D').size())\n #print(df_n)\n #print("========================")\n\n\n #インデックスを振り直す\n df_r = df_n.reset_index()\n #print(df_r)\n #print("-----------")\n\n #２列目の全部の行を取得する ⇨ 日付データは取得しない\n y = df_r.iloc[:,1]\n\n f = 4\n s = 8\n t = 5\n\n # n = 157\n n = len(y)\n macd_list = []\n signal_list = []\n histgram_list = []\n\n for i in range(n):\n if i < f:\n macd_list.append(0)\n signal_list.append(0)\n histgram_list.append(0)\n else :\n macd = sum(y[i-f+1:i+1])/len(y[i-f+1:i+1]) - sum(y[max(0,i-s):i+1])/len(y[max(0,i-s):i+1])\n macd_list.append(macd)\n signal = sum(macd_list[max(0,i-t+1):i+1])/len(macd_list[max(0,i-t+1):i+1])\n signal_list.append(signal)\n histgram = macd - signal\n histgram_list.append(histgram)\n\n #len(my) = 157\n my = np.array(histgram_list)\n #print(my)\n\n ave = np.average(my)\n std = np.std(my)\n sikiiti = ave + 3 * std\n #print(sikiiti)\n\n time = []\n for i in range(len(my)):\n if my[i] > sikiiti:\n #print(df_n.index[i].strftime('%Y-%m-%d'))\n time.append(df_n.index[i].strftime('%Y-%m-%d'))\n #print(df_s[df_n.index[i].strftime('%Y/%m/%d'):df_n.index[i].strftime('%Y/%m/%d')])\n else:\n pass\n\n #print(time)\n\n # len(time) = 4 #\n #C_score_norm =[]\n C_score_norm =[0]*len(df_i)\n\n if len(time) == 0:\n for i in range(len(df_s)):\n C_score_norm.append(0)\n else:\n for h in range(len(time)):\n date = time[h]\n #print(date)\n\n #df_iから対象となるdateのデータを取得する\n aaa = df_i[f'{date} 00:00:00':f'{date} 23:59:59']\n #print(aaa)\n\n # 15T = 15分ごと\n # aaaを15分ごとに刻む\n df_n = pd.DataFrame(aaa.resample('15T').size())\n #print(df_n)\n\n # df_nのインデックスを振り直す\n df_r = df_n.reset_index()\n\n # df_rから評価値の部分だけ取得する\n y = df_r.iloc[:,1]\n #print(y)\n\n n = len(y)\n #print(n)\n macd_list = []\n signal_list = []\n histgram_list = []\n\n for i in range(n):\n if i < f:\n macd_list.append(0)\n signal_list.append(0)\n histgram_list.append(0)\n else :\n macd = sum(y[i-f+1:i+1])/len(y[i-f+1:i+1]) - sum(y[max(0,i-s):i+1])/len(y[max(0,i-s):i+1])\n macd_list.append(macd)\n signal = sum(macd_list[max(0,i-t+1):i+1])/len(macd_list[max(0,i-t+1):i+1])\n signal_list.append(signal)\n histgram = macd - signal\n histgram_list.append(histgram)\n my = np.array(histgram_list)\n\n ave = np.average(my)\n std = np.std(my)\n sikiiti = ave + 3 * std\n\n #print(sikiiti)\n\n target_time = []\n\n #print(my)\n\n for i in range(len(my)):\n if my[i] > sikiiti:\n target_time.append(df_n.index[i].strftime('%Y-%m-%d %H:%M:%S'))\n #print(target_time)\n else:\n pass\n\n #print(len(target_time))\n\n time_new = []\n size = []\n C_score = []\n\n for i in range(len(target_time)):\n time_new.append(datetime.datetime.strptime(target_time[i], '%Y-%m-%d %H:%M:%S'))#datetime型に変換\n size.append(len(df_i[f'{time_new[i]}':f'{time_new[i]+datetime.timedelta(minutes=15)}']))\n C_score.append(math.log(size[i]))\n\n C_score_max = max(C_score, default = C_score)\n #print(C_score)\n\n #print(len(time_new))\n if len(time_new) >= 1:\n for j in range(len(time_new)):\n for i in range(len(df_i)):\n if time_new[j] < df_i.index[i] < time_new[j]+datetime.timedelta(minutes=15):\n #C_score_norm.append(5 * C_score[j] / C_score_max)\n C_score_norm[i] = (5 * C_score[j] / C_score_max)\n else:\n #C_score_norm.append(0)\n C_score_norm[i] = 0\n else:\n pass\n\n return C_score_norm\n\n###########################\n\n#信頼性スコアの計算（HP教材版）\ndef Credibility(S_score_norm,I_score_norm,C_score_norm,HPnumber,df):\n F_score = []\n index = []\n #print(df.query(f'Hpnumber == {HPnumber}').index)\n index = df.query(f'Hpnumber == {HPnumber}').index\n sum = 0\n for i in range(len(index)):\n F_score.append*1:\n F_score.append((S_score_norm[index[i]]+I_scor e_norm[index[i]]+C_score_norm[index[i]])/3)\n sum += df.at[index[i],'evaluation']\n Ave_score = sum/len(index)\n #print(F_score)\n #print(Ave_score)\n #print(statistics.mean(F_score))\n if statistics.mean(F_score) == 0:\n F_cre_score = 0\n else:\n F_cre_score = Ave_score/statistics.mean(F_score)\n #print(F_cre_score)\n\n return F_cre_score,Ave_score\n\n#csvの更新(これはEigoBunpouのみなので，app.pyの方で教科に応じて対象のcsvを変数でおき，その変数を用いてcsvに上書きをする．また，437行目の部分も変数で指定する．)\n#使用するcsvは同一ファイル上にあるから，指定できるはず\n#kyouzaiDB ⇨ kyouzai.csv\n)
nsga2まとめ?
メモ帳
チェッカー