Tutorial 1 — Getting Started¶

This notebook verifies your metasignal installation and walks through the three input arrays every function expects.

1. Verify the install¶

In [1]:

import numpy as np
import metasignal
from metasignal import stdpy

print(f"metasignal loaded from: {metasignal.__file__}")

rng  = np.random.default_rng(0)
stim = rng.choice([0, 1], 200)
resp = np.where(rng.random(200) < 0.75, stim, 1 - stim)
conf = rng.integers(1, 5, 200)

meas = stdpy.compute_all_measures(stim, resp, conf, n_ratings=4)
print("Output shape:", meas.shape)   # (26,)

import numpy as np import metasignal from metasignal import stdpy print(f"metasignal loaded from: {metasignal.__file__}") rng = np.random.default_rng(0) stim = rng.choice([0, 1], 200) resp = np.where(rng.random(200) < 0.75, stim, 1 - stim) conf = rng.integers(1, 5, 200) meas = stdpy.compute_all_measures(stim, resp, conf, n_ratings=4) print("Output shape:", meas.shape) # (26,)

metasignal loaded from: /Users/saurabhext/Documents/metasignal/src/metasignal/__init__.py

Output shape: (26,)

2. Input format¶

Argument	Values	Meaning
stim	0 / 1	Stimulus category (0 = S1/noise, 1 = S2/signal)
resp	0 / 1	Participant response
conf	1 … n_ratings	Confidence rating (1 = lowest)
n_ratings	int	Total number of confidence categories

3. Build a minimal dataset¶

In [2]:

rng = np.random.default_rng(42)
n_trials, n_ratings = 300, 4

stim = rng.choice([0, 1], n_trials)
resp = np.where(rng.random(n_trials) < 0.80, stim, 1 - stim)
correct = (stim == resp)
conf = np.where(
    correct,
    rng.integers(3, n_ratings + 1, n_trials),
    rng.integers(1, 3, n_trials),
)

print(f"Trials   : {n_trials}")
print(f"Accuracy : {correct.mean():.1%}")
print(f"Mean conf: {conf.mean():.2f}")

rng = np.random.default_rng(42) n_trials, n_ratings = 300, 4 stim = rng.choice([0, 1], n_trials) resp = np.where(rng.random(n_trials) < 0.80, stim, 1 - stim) correct = (stim == resp) conf = np.where( correct, rng.integers(3, n_ratings + 1, n_trials), rng.integers(1, 3, n_trials), ) print(f"Trials : {n_trials}") print(f"Accuracy : {correct.mean():.1%}") print(f"Mean conf: {conf.mean():.2f}")

Trials   : 300
Accuracy : 77.3%
Mean conf: 3.05

4. Type-1 SDT parameters¶

In [3]:

dprime, c, ln_beta = stdpy.compute_sdt_resp(stim, resp)
print(f"d'        = {dprime:.3f}")
print(f"criterion = {c:.3f}")

dprime, c, ln_beta = stdpy.compute_sdt_resp(stim, resp) print(f"d' = {dprime:.3f}") print(f"criterion = {c:.3f}")

d'        = 1.499
criterion = 0.069

5. Convert trials to response counts¶

In [4]:

nr_s1, nr_s2 = stdpy.trials_to_counts(stim, resp, conf, n_ratings=n_ratings)
print("nr_s1 shape:", nr_s1.shape)
print("nr_s1:", nr_s1)
print("nr_s2:", nr_s2)

nr_s1, nr_s2 = stdpy.trials_to_counts(stim, resp, conf, n_ratings=n_ratings) print("nr_s1 shape:", nr_s1.shape) print("nr_s1:", nr_s1) print("nr_s2:", nr_s2)

nr_s1 shape: (8,)
nr_s1: [60. 63.  0.  0. 16. 16.  0.  0.]
nr_s2: [ 0.  0. 20. 16.  0.  0. 53. 56.]

6. Inspect the full 26-element output¶

In [5]:

MEASURE_NAMES = [
    "meta-d'", "AUC2", "Gamma", "Phi", "DeltaConf",
    "M-Ratio", "AUC2-Ratio", "Gamma-Ratio", "Phi-Ratio", "DeltaConf-Ratio",
    "M-Diff", "AUC2-Diff", "Gamma-Diff", "Phi-Diff", "DeltaConf-Diff",
    "metaNoise", "metaUncertainty", "d'", "c", "mean_conf",
    "logL", "AIC", "BIC", "AICc", "k", "n",
]

meas = stdpy.compute_all_measures(stim, resp, conf, n_ratings=n_ratings)

for i, (name, val) in enumerate(zip(MEASURE_NAMES, meas)):
    flag = "NaN" if np.isnan(val) else f"{val:.4f}"
    print(f"  [{i:2d}] {name:<20s} = {flag}")

MEASURE_NAMES = [ "meta-d'", "AUC2", "Gamma", "Phi", "DeltaConf", "M-Ratio", "AUC2-Ratio", "Gamma-Ratio", "Phi-Ratio", "DeltaConf-Ratio", "M-Diff", "AUC2-Diff", "Gamma-Diff", "Phi-Diff", "DeltaConf-Diff", "metaNoise", "metaUncertainty", "d'", "c", "mean_conf", "logL", "AIC", "BIC", "AICc", "k", "n", ] meas = stdpy.compute_all_measures(stim, resp, conf, n_ratings=n_ratings) for i, (name, val) in enumerate(zip(MEASURE_NAMES, meas)): flag = "NaN" if np.isnan(val) else f"{val:.4f}" print(f" [{i:2d}] {name:<20s} = {flag}")

  [ 0] meta-d'              = 7.6642
  [ 1] AUC2                 = 1.0000
  [ 2] Gamma                = 1.0000
  [ 3] Phi                  = 0.8553
  [ 4] DeltaConf            = 1.9706
  [ 5] M-Ratio              = 5.1551
  [ 6] AUC2-Ratio           = 1.4269
  [ 7] Gamma-Ratio          = 1.6859
  [ 8] Phi-Ratio            = 2.7669
  [ 9] DeltaConf-Ratio      = 2.9124
  [10] M-Diff               = 6.1775
  [11] AUC2-Diff            = 0.2992
  [12] Gamma-Diff           = 0.4068
  [13] Phi-Diff             = 0.5462
  [14] DeltaConf-Diff       = 1.2940
  [15] metaNoise            = 0.1008
  [16] metaUncertainty      = 0.0100
  [17] d'                   = 1.4987
  [18] c                    = 0.0694
  [19] mean_conf            = 3.0533
  [20] logL                 = -220.3121
  [21] AIC                  = 454.6242
  [22] BIC                  = 480.5972
  [23] AICc                 = 455.0051
  [24] k                    = 7.0000
  [25] n                    = 302.0000