はじめての 統計データ分析 ―ベイズ的〈ポストp値時代〉の統計学― その5
その5です。今回は第4章の章末問題に取り組んでいきます。
4章 対応ある2群の差と相関の推測
内容
- 対応ある2群のt検定のオルタナとして機能します
- 対応ある、の意味とは?
- 同じ観察対象から2回測定しているもの
- beforeの体重とafterの体重のセット * n個など
- この解析をするときには実験デザインが大事になります
- どちらかの群にバイアスがかかることを避ける事が必要です
- 対応ある2群の実験デザインを行う
- マッチング : 施策実行前の状態が同じ2つを組にして、ランダムに2群に割り当てる
- プリテスト・ポストテスト : 施策の前後で同じ対象を観察する
- 相関関係を表現する要約統計量の例
- 共分散
- 平均偏差の積の平均値
- 正の相関がある時に正、負の相関がある時に負の値になる
- 相関の強さは表現できない
- 相関係数 or
- データの正規化を行った後に、積の平均値を計算
- -1 <= r <= 1
- 共分散
- 2変量正規分布の導入
- 共分散: の関係性を持つ
- 正規分布に従う2変量が観測される確率をモデル化する際に、あてはめが可能となる理論分布
- 2群の差異の考察のバリエーション
- 独立した2群の差の分析
- 対応ある2群の群間差の分析(Inter)
- 対応ある2群の個人内差の分析(Intra)
- 対応ある場合の生成量(2変量に相関が無い場合は、とすればokです
章末問題
データの取得
取り敢えずデータを準備します。
import numpy as np import os from logging import getLogger, Formatter, StreamHandler, DEBUG # printではなくloggerを使う def get_logger(): logger = getLogger(__name__) logger.setLevel(DEBUG) log_fmt = '%(asctime)s : %(name)s : %(levelname)s : %(message)s' formatter = Formatter(log_fmt) stream_handler = StreamHandler() stream_handler.setLevel(DEBUG) stream_handler.setFormatter(formatter) logger.addHandler(stream_handler) return logger logger = get_logger() # データの準備 a = np.array([62,54,19,54,47,22,35,77,64,60,27,41,41,44,57,16,42,89,40,67,69,46,74,62,60,87,32,42,73,25,42,57,31,35,33,38,43,53,55,62,67,56,76,5,31,70,66,65,34,48]) b = np.array([73,72,56,58,71,42,78,77,75,72,56,71,69,77,84,51,62,88,56,58,84,91,71,82,81,77,65,78,79,60,66,70,65,57,64,61,56,67,75,64,68,67,80,55,48,85,56,62,65,79]) x = np.stack((a, b), axis=1) # 出力用ディレクトリの用意 result_dir = os.path.join("result", "chapter4") if os.path.exists(result_dir) is False: os.makedirs(result_dir) logger.info("{0} is made".format(result_dir))
各種統計量の計算と可視化
モデルから推定をする前に、統計量を計算していきます。 対応関係を使って散布図を書けば相関が見えてきますね。
import pandas as pd import matplotlib.pyplot as plt import seaborn as sns from tabulate import tabulate from IPython.core.display import display # データの要約統計量を計算 df = pd.DataFrame({"a":a, "b":b}) df_desc = df.describe() display(df_desc) df_desc_path = os.path.join(result_dir, "df_describe.md") with open(df_desc_path, "w") as f: f.write(tabulate(df_desc, df_desc.columns, tablefmt="pipe")) logger.info("sample data summary is saved at {0}".format(df_desc_path)) # 散布図を可視化 sns.jointplot(data=df, x="a", y="b") jointplot_path = os.path.join(result_dir, "jointplot.png") plt.show() plt.savefig(jointplot_path) logger.info("jointplot result is saved at {0}".format(jointplot_path)) r = np.corrcoef(x.T)[0][1] logger.info("ピアソン相関係数は{0}です".format(r)) s = np.cov(x.T)[0][1] logger.info("共分散は{0}です".format(s))
- 結果(表と分布)
a | b | |
---|---|---|
count | 50 | 50 |
mean | 49.9 | 68.48 |
std | 18.6703 | 10.9958 |
min | 5 | 42 |
25% | 35.75 | 60.25 |
50% | 50.5 | 68.5 |
75% | 63.5 | 77 |
max | 89 | 91 |
Stanによる統計モデルの構築
これもしょうもないテクニックなんですが、2次元配列 or 行列がパラメータとして得られている場合には1次元の配列に入れています。PyStanの可視化部分が多次元構造に対応しておらず、突然カーネルが死んでしまうためです(summaryメソッドは対応しています)。PyStanの可視化は昔のPyMC3のモジュールを利用しているのですが、完璧に対応できていないことが原因のようです。その内PyMC3のチームが作っているらしいmcmcplotlibというモジュールに移行する予定らしいですが、まだその雰囲気はありません…
また同順率を計算する時にsinの逆関数としてasinを利用しました。
import os import pystan import pickle # Stanのモデルを読み込んでコンパイルする stan_file = os.path.join("stan", "g2_pair.stan") stan_file_c = os.path.join("stan", "g2_pair.pkl") model = pystan.StanModel(file=stan_file) with open(stan_file_c, "wb") as f: pickle.dump(model, f) logger.info("Stan model is compiled to {0}".format(stan_file_c))
- Stan
data { int<lower=0> n ; vector[2] x[n] ; real c_mu_diff ; real c_es ; real c_cohenu ; real c_pod ; real c_pbt ; real<lower=0, upper=1> cdash_pbt ; real c_diff_sd ; real c_pair_es ; real<lower=0, upper=1> c_pair_pod ; real<lower=0, upper=1> cdash_pair_pbt ; real c_rho ; real<lower=0, upper=1> c_poc; } parameters { vector[2] mu ; real<lower=0> sigma_a ; real<lower=0> sigma_b ; real<lower=-1, upper=1> rho ; } transformed parameters { # ダイレクトに共分散行列を与えると # PyStanの可視化でエラー無しに落ちるので、Stan内部で作る real<lower=0> sigma_ab ; cov_matrix[2] Sigma ; sigma_ab = sigma_a * sigma_b * rho ; Sigma[1,1] = pow(sigma_a, 2) ; Sigma[1,2] = sigma_ab ; Sigma[2,1] = sigma_ab ; Sigma[2,2] = pow(sigma_b, 2) ; } model { for(i in 1:n){ x[i] ~ multi_normal(mu, Sigma) ; } } generated quantities { vector[n] log_lik ; real mu_diff ; real es_a ; real es_b ; real cohenu_a ; real cohenu_b ; real<lower=0, upper=1> pod ; real<lower=0, upper=1> pbt ; real diff_sd ; real pair_es ; real pair_pod ; real pair_pbt ; real<lower=0, upper=1> poc ; int<lower=0, upper=1> prob_mu_diff_upper_0 ; int<lower=0, upper=1> prob_mu_diff_upper_c ; int<lower=0, upper=1> prob_es_a_upper_c ; int<lower=0, upper=1> prob_es_b_upper_c ; int<lower=0, upper=1> prob_cohenu_a_upper_c ; int<lower=0, upper=1> prob_pod_upper_c ; int<lower=0, upper=1> prob_pbt_upper_cdash ; int<lower=0, upper=1> prob_diff_sd_upper_c ; int<lower=0, upper=1> prob_pair_es_upper_c ; int<lower=0, upper=1> prob_pair_pod_upper_c ; int<lower=0, upper=1> prob_pair_pbt_upper_cdash ; int<lower=0, upper=1> prob_rho_upper_c ; int<lower=0, upper=1> prob_poc_upper_c ; for(i in 1:n){ log_lik[i] = multi_normal_lpdf(x[i] | mu, Sigma) ; } mu_diff = mu[1] - mu[2] ; es_a = fabs(mu[1] - mu[2]) / sigma_a ; es_b = fabs(mu[1] - mu[2]) / sigma_b ; cohenu_a = normal_cdf(mu[2], mu[1], sigma_a) ; cohenu_b = normal_cdf(mu[1], mu[2], sigma_b) ; pod = normal_cdf(mu_diff / sqrt(pow(sigma_a, 2) + pow(sigma_b, 2)), 0, 1) ; pbt = normal_cdf((mu_diff - c_pbt) / sqrt(pow(sigma_a, 2) + pow(sigma_b, 2)), 0, 1) ; diff_sd = sqrt(pow(sigma_a, 2) + pow(sigma_b, 2) - 2 * rho * sigma_a * sigma_b) ; pair_es = fabs(mu_diff) / diff_sd; pair_pod = normal_cdf(pair_es, 0, 1) ; pair_pbt = normal_cdf((mu_diff - c_pbt) / diff_sd, 0, 1) ; poc = 0.5 + 1 / pi() * asin(rho) ; prob_mu_diff_upper_0 = mu_diff > 0 ? 1 : 0 ; prob_mu_diff_upper_c = mu_diff > c_mu_diff ? 1 : 0 ; prob_es_a_upper_c = es_a > c_es ? 1 : 0 ; prob_es_b_upper_c = es_b > c_es ? 1 : 0 ; prob_cohenu_a_upper_c = cohenu_a > c_cohenu ? 1 : 0 ; prob_pod_upper_c = pod > c_pod ? 1 : 0 ; prob_pbt_upper_cdash = pbt > cdash_pbt ? 1 : 0 ; prob_diff_sd_upper_c = diff_sd > c_diff_sd ? 1 : 0 ; prob_pair_es_upper_c = pair_es > c_pair_es ? 1 : 0 ; prob_pair_pod_upper_c = pair_pod > c_pair_pod ? 1 : 0 ; prob_pair_pbt_upper_cdash = pair_pbt > cdash_pair_pbt ? 1 : 0 ; prob_rho_upper_c = rho > c_rho ? 1 : 0 ; prob_poc_upper_c = poc > c_poc ? 1 : 0 ; }
統計モデルによる事後分布のサンプリング
prob_mu_diff_upper_0については全てのサンプリング値で0となり、pystanのモジュールで可視化出来ない(カーネル密度推定出来ない)のでその直前で除きます。
import pandas as pd import pickle import pystan import matplotlib import matplotlib.pyplot as plt import os from IPython.core.display import display from tabulate import tabulate matplotlib.rcParams['figure.figsize'] = (10, 80) # Stanで使うデータの用意 stan_data = {"n": len(x), "x": x, "c_mu_diff": -15, "c_es": 1.0, "c_cohenu": 0.8, "c_pod": 0.1, "c_pbt": 15, "cdash_pbt": 0.05, "c_diff_sd": 20, "c_pair_es": 1.0, "c_pair_pod": 0.8, "cdash_pair_pbt": 0.05, "c_rho": 0.75, "c_poc": 0.75} # 興味のあるパラメータの設定 par = ["mu", "sigma_a", "sigma_b", "rho", "sigma_ab", "mu_diff", "es_a", "es_b", "cohenu_a", "cohenu_b", "pod", "pbt", "diff_sd", "pair_es", "pair_pod", "pair_pbt", "poc", "prob_mu_diff_upper_0", "prob_mu_diff_upper_c", "prob_es_a_upper_c", "prob_es_b_upper_c", "prob_cohenu_a_upper_c", "prob_pod_upper_c", "prob_pbt_upper_cdash", "prob_diff_sd_upper_c", "prob_pair_es_upper_c", "prob_pair_pod_upper_c", "prob_pair_pbt_upper_cdash", "prob_rho_upper_c", "prob_poc_upper_c", "log_lik"] prob = [0.025, 0.05, 0.25, 0.5, 0.75, 0.95, 0.975] # モデルの読み込み stan_file_c = os.path.join("stan", "g2_pair.pkl") with open(stan_file_c, "rb") as f: model = pickle.load(f) # MCMCでサンプリング logger.info("Start MCMC sampling") fit = model.sampling(data=stan_data, pars=par, iter=21000, chains=5, warmup=1000, seed=1234, algorithm="NUTS") # 事後分布の表を取得 summary = fit.summary(pars=par, probs=prob) summary_df = pd.DataFrame(summary["summary"], index=summary["summary_rownames"], columns=summary["summary_colnames"]) display(summary_df) summary_df_path = os.path.join(result_dir, "df_summary.md") with open(summary_df_path, "w") as f: f.write(tabulate(summary_df, summary_df.columns, tablefmt="pipe")) logger.info("MCMC result summary is saved at {0}".format(summary_df_path)) # 事後分布の可視化 par.remove("prob_mu_diff_upper_0") fit.traceplot(par) traceplot_path = os.path.join(result_dir, "traceplot.png") plt.savefig(traceplot_path) plt.show() logger.info("traceplot result is saved at {0}".format(traceplot_path)) # WAICの計算 log_lik = fit.extract("log_lik")["log_lik"] waic = -2 * np.sum(np.log(np.mean(np.exp(log_lik), axis=0))) + 2 * np.sum(np.var(log_lik, axis=0)) logger.info("WAICの値は{0}です".format(waic))
- 結果(表と分布)
mean | se_mean | sd | 2.5% | 5% | 25% | 50% | 75% | 95% | 97.5% | n_eff | Rhat | |
---|---|---|---|---|---|---|---|---|---|---|---|---|
mu[0] | 49.8926 | 0.00957801 | 2.73487 | 44.4876 | 45.3974 | 48.0818 | 49.8909 | 51.7234 | 54.3679 | 55.2544 | 81531 | 1.00003 |
mu[1] | 68.4765 | 0.00566431 | 1.60689 | 65.3396 | 65.8384 | 67.4029 | 68.4804 | 69.5427 | 71.1116 | 71.6325 | 80478 | 1.00001 |
sigma_a | 19.2229 | 0.00716584 | 2.00791 | 15.7698 | 16.2345 | 17.8062 | 19.061 | 20.45 | 22.7575 | 23.6287 | 78515 | 0.999995 |
sigma_b | 11.3205 | 0.00418226 | 1.18074 | 9.2944 | 9.56625 | 10.4895 | 11.2246 | 12.0418 | 13.4129 | 13.9161 | 79705 | 0.999964 |
rho | 0.594667 | 0.000335941 | 0.0935382 | 0.390446 | 0.428657 | 0.536483 | 0.602234 | 0.661714 | 0.733422 | 0.754116 | 77527 | 0.999996 |
sigma_ab | 131.615 | 0.153909 | 38.1313 | 69.8853 | 77.6958 | 104.862 | 127.231 | 153.284 | 200.727 | 219.867 | 61381 | 0.999985 |
mu_diff | -18.5839 | 0.00693064 | 2.19166 | -22.9048 | -22.1834 | -20.0325 | -18.5904 | -17.1309 | -14.9834 | -14.2762 | 100000 | 1.00003 |
es_a | 0.977031 | 0.000478766 | 0.151399 | 0.687624 | 0.732986 | 0.873885 | 0.97449 | 1.07755 | 1.22976 | 1.27962 | 100000 | 1.00005 |
es_b | 1.65908 | 0.000820216 | 0.259375 | 1.18273 | 1.25308 | 1.47896 | 1.64867 | 1.82674 | 2.10267 | 2.19663 | 100000 | 1 |
cohenu_a | 0.832983 | 0.000118209 | 0.0373811 | 0.754155 | 0.768216 | 0.808909 | 0.835093 | 0.859383 | 0.890606 | 0.89966 | 100000 | 1.00004 |
cohenu_b | 0.0540274 | 8.64856e-05 | 0.0273491 | 0.0140236 | 0.0177475 | 0.0338697 | 0.0496074 | 0.0695757 | 0.105088 | 0.118457 | 100000 | 0.999984 |
pod | 0.202597 | 0.000108955 | 0.0344545 | 0.139349 | 0.148333 | 0.178551 | 0.201236 | 0.225017 | 0.261581 | 0.273534 | 100000 | 1.00004 |
pbt | 0.0674462 | 7.64904e-05 | 0.0216195 | 0.0324478 | 0.0364988 | 0.0518413 | 0.065037 | 0.0803561 | 0.10656 | 0.116507 | 79887 | 1.00001 |
diff_sd | 15.4018 | 0.00515225 | 1.62928 | 12.6101 | 12.9776 | 14.2577 | 15.2707 | 16.3993 | 18.2913 | 18.9678 | 100000 | 0.999976 |
pair_es | 1.21977 | 0.000598837 | 0.189369 | 0.851131 | 0.910184 | 1.09139 | 1.21844 | 1.34708 | 1.5329 | 1.59392 | 100000 | 1.00002 |
pair_pod | 0.884636 | 0.00011549 | 0.0365212 | 0.802652 | 0.818637 | 0.86245 | 0.888472 | 0.911022 | 0.93735 | 0.944523 | 100000 | 1.00001 |
pair_pbt | 0.0165938 | 3.57684e-05 | 0.0113109 | 0.00309664 | 0.00399766 | 0.0085756 | 0.013873 | 0.0216362 | 0.0384226 | 0.0455154 | 100000 | 0.999988 |
poc | 0.704289 | 0.000133159 | 0.0369491 | 0.627679 | 0.641013 | 0.680247 | 0.705723 | 0.730171 | 0.762078 | 0.771934 | 76996 | 0.999993 |
prob_mu_diff_upper_0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 100000 | nan |
prob_mu_diff_upper_c | 0.05073 | 0.000763322 | 0.219447 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 82650 | 0.999997 |
prob_es_a_upper_c | 0.43221 | 0.00156655 | 0.495386 | 0 | 0 | 0 | 0 | 1 | 1 | 1 | 100000 | 1.00004 |
prob_es_b_upper_c | 0.99731 | 0.000176451 | 0.0517957 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 86167 | 1.00002 |
prob_cohenu_a_upper_c | 0.81293 | 0.00136484 | 0.38997 | 0 | 0 | 1 | 1 | 1 | 1 | 1 | 81639 | 0.999982 |
prob_pod_upper_c | 0.99972 | 5.65467e-05 | 0.0167309 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 87544 | 1.00006 |
prob_pbt_upper_cdash | 0.78259 | 0.0013449 | 0.412486 | 0 | 0 | 1 | 1 | 1 | 1 | 1 | 94067 | 1.00003 |
prob_diff_sd_upper_c | 0.00844 | 0.00033347 | 0.0914814 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 75258 | 1.00001 |
prob_pair_es_upper_c | 0.87754 | 0.00116844 | 0.327818 | 0 | 0 | 1 | 1 | 1 | 1 | 1 | 78714 | 0.999991 |
prob_pair_pod_upper_c | 0.97798 | 0.000521882 | 0.146749 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 79069 | 0.999996 |
prob_pair_pbt_upper_cdash | 0.01667 | 0.000463955 | 0.128032 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 76153 | 1.00003 |
prob_rho_upper_c | 0.0288 | 0.000634729 | 0.167245 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 69427 | 1.00002 |
prob_poc_upper_c | 0.10329 | 0.0011511 | 0.304339 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 69902 | 0.999987 |
log_lik[0] | -7.20677 | 0.000510806 | 0.153203 | -7.51933 | -7.46593 | -7.30743 | -7.20317 | -7.10048 | -6.96204 | -6.9202 | 89954 | 1.00002 |
log_lik[1] | -7.04915 | 0.000502755 | 0.148317 | -7.35091 | -7.29856 | -7.14718 | -7.04574 | -6.94693 | -6.81087 | -6.76865 | 87030 | 0.999998 |
log_lik[2] | -8.37423 | 0.0010673 | 0.313067 | -9.05396 | -8.9247 | -8.56952 | -8.3508 | -8.15363 | -7.90214 | -7.82931 | 86041 | 0.999984 |
log_lik[3] | -7.93282 | 0.000748297 | 0.236632 | -8.43927 | -8.34739 | -8.0812 | -7.91782 | -7.76714 | -7.57149 | -7.51186 | 100000 | 1.00001 |
log_lik[4] | -7.09005 | 0.000503034 | 0.148439 | -7.39144 | -7.3404 | -7.18795 | -7.08658 | -6.98816 | -6.85275 | -6.80871 | 87077 | 1 |
log_lik[5] | -9.88101 | 0.00200767 | 0.594664 | -11.1672 | -10.9272 | -10.2558 | -9.83603 | -9.45866 | -8.98453 | -8.85144 | 87732 | 0.999988 |
log_lik[6] | -8.71399 | 0.00119432 | 0.377678 | -9.53705 | -9.38041 | -8.9495 | -8.68412 | -8.44366 | -8.15071 | -8.06613 | 100000 | 1.00001 |
log_lik[7] | -8.05166 | 0.000808855 | 0.255782 | -8.60068 | -8.49872 | -8.2127 | -8.03498 | -7.87169 | -7.66212 | -7.60015 | 100000 | 1.00006 |
log_lik[8] | -7.29648 | 0.000531128 | 0.158762 | -7.62103 | -7.56603 | -7.40109 | -7.29165 | -7.18636 | -7.04438 | -7.00056 | 89350 | 1.00003 |
log_lik[9] | -7.14304 | 0.000503928 | 0.150479 | -7.44972 | -7.39716 | -7.24229 | -7.13974 | -7.03927 | -6.90195 | -6.85998 | 89169 | 1.00001 |
log_lik[10] | -7.86887 | 0.000793813 | 0.227803 | -8.35733 | -8.26465 | -8.01223 | -7.85569 | -7.70969 | -7.51842 | -7.4618 | 82354 | 0.999979 |
log_lik[11] | -7.31585 | 0.000503531 | 0.15923 | -7.6417 | -7.58604 | -7.41964 | -7.31028 | -7.20657 | -7.06351 | -7.01762 | 100000 | 1.00002 |
log_lik[12] | -7.19628 | 0.000480615 | 0.151984 | -7.50551 | -7.45265 | -7.29581 | -7.19238 | -7.09229 | -6.95323 | -6.90778 | 100000 | 1.00001 |
log_lik[13] | -7.76793 | 0.000671836 | 0.212453 | -8.21805 | -8.13685 | -7.90243 | -7.75448 | -7.61944 | -7.44257 | -7.38864 | 100000 | 1.00003 |
log_lik[14] | -8.15601 | 0.000870612 | 0.275312 | -8.75044 | -8.63974 | -8.32869 | -8.137 | -7.96147 | -7.7388 | -7.67263 | 100000 | 1 |
log_lik[15] | -8.83067 | 0.0014309 | 0.395669 | -9.69151 | -9.52681 | -9.07889 | -8.79907 | -8.55174 | -8.23532 | -8.14442 | 76462 | 0.999986 |
log_lik[16] | -7.17365 | 0.000504963 | 0.151681 | -7.4807 | -7.43007 | -7.27362 | -7.1703 | -7.06917 | -6.93037 | -6.88726 | 90228 | 0.999968 |
log_lik[17] | -9.38805 | 0.00180669 | 0.499185 | -10.4735 | -10.2703 | -9.70282 | -9.35085 | -9.03296 | -8.63428 | -8.52274 | 76341 | 1.00004 |
log_lik[18] | -7.6546 | 0.000618485 | 0.195582 | -8.06579 | -7.99312 | -7.77964 | -7.64522 | -7.51888 | -7.34957 | -7.29857 | 100000 | 0.999987 |
log_lik[19] | -9.16164 | 0.00145992 | 0.461667 | -10.1567 | -9.98055 | -9.45124 | -9.12617 | -8.83245 | -8.47093 | -8.36446 | 100000 | 1 |
log_lik[20] | -8.01269 | 0.00084993 | 0.250002 | -8.54689 | -8.45199 | -8.17131 | -7.99616 | -7.83703 | -7.6313 | -7.57052 | 86521 | 1.00001 |
log_lik[21] | -10.6807 | 0.00242399 | 0.766533 | -12.3495 | -12.0334 | -11.1564 | -10.6244 | -10.1335 | -9.53191 | -9.35498 | 100000 | 0.999991 |
log_lik[22] | -8.05483 | 0.000812016 | 0.256782 | -8.609 | -8.50441 | -8.21509 | -8.03812 | -7.87417 | -7.6654 | -7.60062 | 100000 | 1.00004 |
log_lik[23] | -7.7555 | 0.000666349 | 0.210718 | -8.20175 | -8.11848 | -7.89039 | -7.74328 | -7.6084 | -7.43055 | -7.37751 | 100000 | 1.00001 |
log_lik[24] | -7.65779 | 0.00062335 | 0.19712 | -8.07091 | -7.99542 | -7.78414 | -7.64716 | -7.52059 | -7.35168 | -7.30084 | 100000 | 1.00001 |
log_lik[25] | -9.09361 | 0.00140706 | 0.444952 | -10.0591 | -9.8797 | -9.37279 | -9.05784 | -8.77697 | -8.42635 | -8.32266 | 100000 | 1.00004 |
log_lik[26] | -7.50563 | 0.000563826 | 0.178297 | -7.87608 | -7.81074 | -7.6211 | -7.49828 | -7.38201 | -7.22696 | -7.17773 | 100000 | 0.999996 |
log_lik[27] | -8.05778 | 0.000818254 | 0.258755 | -8.61498 | -8.51212 | -8.22005 | -8.03921 | -7.87442 | -7.66727 | -7.60576 | 100000 | 1.00003 |
log_lik[28] | -7.79465 | 0.000734035 | 0.215728 | -8.25129 | -8.16903 | -7.93286 | -7.78268 | -7.64341 | -7.46225 | -7.40685 | 86373 | 1.00006 |
log_lik[29] | -7.88166 | 0.000749915 | 0.230262 | -8.37495 | -8.28305 | -8.02686 | -7.86715 | -7.72116 | -7.52979 | -7.47299 | 94280 | 0.999984 |
log_lik[30] | -7.08697 | 0.00050002 | 0.148871 | -7.38986 | -7.33785 | -7.18472 | -7.08402 | -6.98546 | -6.84782 | -6.80542 | 88643 | 0.999981 |
log_lik[31] | -7.0756 | 0.000500711 | 0.148709 | -7.37755 | -7.32674 | -7.17359 | -7.072 | -6.97293 | -6.83728 | -6.79434 | 88206 | 0.999993 |
log_lik[32] | -7.57171 | 0.000588542 | 0.186113 | -7.9608 | -7.89077 | -7.69187 | -7.56353 | -7.44214 | -7.28218 | -7.23107 | 100000 | 0.999997 |
log_lik[33] | -7.56191 | 0.000618204 | 0.184662 | -7.94683 | -7.87941 | -7.68109 | -7.55408 | -7.43403 | -7.27271 | -7.22362 | 89226 | 0.999976 |
log_lik[34] | -7.41856 | 0.000535287 | 0.169273 | -7.76974 | -7.70685 | -7.52824 | -7.41221 | -7.30165 | -7.15134 | -7.10458 | 100000 | 0.999989 |
log_lik[35] | -7.26815 | 0.000523735 | 0.157009 | -7.58864 | -7.53398 | -7.37085 | -7.26407 | -7.15992 | -7.01727 | -6.97304 | 89873 | 0.999969 |
log_lik[36] | -7.71186 | 0.00064242 | 0.203151 | -8.14296 | -8.06432 | -7.84074 | -7.70171 | -7.57035 | -7.39709 | -7.34306 | 100000 | 0.999994 |
log_lik[37] | -7.05357 | 0.000500421 | 0.148306 | -7.35433 | -7.30301 | -7.15194 | -7.05027 | -6.95145 | -6.81583 | -6.77401 | 87831 | 0.999979 |
log_lik[38] | -7.17771 | 0.000505919 | 0.151559 | -7.48541 | -7.43305 | -7.278 | -7.17341 | -7.07353 | -6.93566 | -6.89261 | 89743 | 1.00001 |
log_lik[39] | -7.69658 | 0.000637628 | 0.201636 | -8.12377 | -8.04458 | -7.82498 | -7.68571 | -7.55632 | -7.38327 | -7.33149 | 100000 | 1.00001 |
log_lik[40] | -7.6861 | 0.000633419 | 0.200305 | -8.11 | -8.03307 | -7.81326 | -7.67564 | -7.54736 | -7.37526 | -7.32353 | 100000 | 1.00002 |
log_lik[41] | -7.135 | 0.000501372 | 0.14986 | -7.43817 | -7.3869 | -7.2345 | -7.13109 | -7.03222 | -6.89567 | -6.85188 | 89341 | 0.999986 |
log_lik[42] | -8.00336 | 0.000847046 | 0.247836 | -8.53242 | -8.435 | -8.16072 | -7.98708 | -7.82881 | -7.6244 | -7.56385 | 85608 | 1.00006 |
log_lik[43] | -9.90667 | 0.00191457 | 0.605442 | -11.2268 | -10.9773 | -10.2848 | -9.8613 | -9.47535 | -8.99769 | -8.86278 | 100000 | 0.999986 |
log_lik[44] | -8.72917 | 0.00118485 | 0.374681 | -9.5436 | -9.38778 | -8.96398 | -8.70135 | -8.46409 | -8.16497 | -8.07738 | 100000 | 0.999993 |
log_lik[45] | -8.14419 | 0.000925485 | 0.271828 | -8.72857 | -8.62312 | -8.31587 | -8.12515 | -7.95293 | -7.73213 | -7.66608 | 86268 | 1 |
log_lik[46] | -9.48036 | 0.00165699 | 0.523985 | -10.6101 | -10.409 | -9.80909 | -9.43988 | -9.10587 | -8.69649 | -8.57541 | 100000 | 1 |
log_lik[47] | -8.21578 | 0.000900439 | 0.284744 | -8.82678 | -8.71809 | -8.39352 | -8.19606 | -8.01586 | -7.78551 | -7.7145 | 100000 | 1.00001 |
log_lik[48] | -7.38682 | 0.000525221 | 0.16609 | -7.7291 | -7.66924 | -7.49486 | -7.38077 | -7.27242 | -7.12421 | -7.07814 | 100000 | 0.999993 |
log_lik[49] | -7.80583 | 0.000690243 | 0.218274 | -8.26971 | -8.18366 | -7.94377 | -7.79197 | -7.65319 | -7.47189 | -7.41726 | 100000 | 1.00003 |