{
  "task_id": "F6-conf-01",
  "n_missing_rows": 215,
  "n_train": 2500,
  "n_test": 2501,
  "anova_f_stat": 0.26346571680711994,
  "anova_p_value": 0.9015060507654888,
  "effect_estimate": 0.05017009840169966,
  "effect_ci_low": -0.319335775694099,
  "effect_ci_high": 0.41967597249749833,
  "baseline_metric": 4.875704154642543,
  "model_metric": 4.883841522290305,
  "conclusion": "no_detectable_effect",
  "feasibility": "complete",
  "feasibility_reason": null,
  "data_issues": [
    {
      "issue": "215 rows with empty DGS10 field (US market holidays / no publication)",
      "handling": "rows dropped before computing changes, as required; the change after a gap spans >1 business day and is attributed to the weekday of its end date"
    },
    {
      "issue": "214 changes span a holiday gap (longer than the usual 1 or 3 calendar days); this mixes horizons across weekdays (e.g. Tuesdays after Monday holidays)",
      "handling": "kept as specified by TASK.md; noted as a limitation"
    }
  ],
  "n_input_rows": {
    "DGS10.csv": 5217
  },
  "n_analysed": 5001,
  "random_seed": null,
  "supplementary": {
    "monday_contrast_method": "Welch two-sample t (Monday vs other weekdays)",
    "welch_se_bp": 0.18837955539096202,
    "welch_df": 1548.9581965107568,
    "welch_p_value": 0.7900247217974673,
    "train_weekday_means_bp": {
      "Monday": -0.05434782608695535,
      "Tuesday": -0.40545808966861635,
      "Wednesday": -0.20542635658914643,
      "Thursday": 0.013833992094862349,
      "Friday": 0.11683168316831462
    },
    "rmse_train_overall_mean_bp": 4.87519865288523,
    "rmse_ratio_model_to_baseline": 1.0016689625518016,
    "n_valid_obs": 5002
  }
}