This vignette demonstrates how drasR can be used with
lot-level inspection data.
In real manufacturing or pharmaceutical inspection settings, users may not know the true defect probability. Instead, they usually observe the number of inspected units and the number of defects found in each lot.
raw_data <- data.frame(
lot_id = paste0("L", 1:12),
date = as.Date("2026-01-01") + 0:11,
supplier = c(rep("Supplier_A", 6), rep("Supplier_B", 6)),
lot_size = rep(1000, 12),
sample_size = rep(100, 12),
defects_found = c(1, 0, 1, 2, 1, 2, 4, 5, 6, 4, 7, 8)
)
raw_data## lot_id date supplier lot_size sample_size defects_found
## 1 L1 2026-01-01 Supplier_A 1000 100 1
## 2 L2 2026-01-02 Supplier_A 1000 100 0
## 3 L3 2026-01-03 Supplier_A 1000 100 1
## 4 L4 2026-01-04 Supplier_A 1000 100 2
## 5 L5 2026-01-05 Supplier_A 1000 100 1
## 6 L6 2026-01-06 Supplier_A 1000 100 2
## 7 L7 2026-01-07 Supplier_B 1000 100 4
## 8 L8 2026-01-08 Supplier_B 1000 100 5
## 9 L9 2026-01-09 Supplier_B 1000 100 6
## 10 L10 2026-01-10 Supplier_B 1000 100 4
## 11 L11 2026-01-11 Supplier_B 1000 100 7
## 12 L12 2026-01-12 Supplier_B 1000 100 8prepared <- prepare_lot_data(
raw_data,
lot_id = "lot_id",
sample_size = "sample_size",
defects = "defects_found",
lot_size = "lot_size",
date = "date",
supplier = "supplier"
)
prepared## lot lot_id sample_size sample_defects lot_size date supplier_id
## 1 1 L1 100 1 1000 2026-01-01 Supplier_A
## 2 2 L2 100 0 1000 2026-01-02 Supplier_A
## 3 3 L3 100 1 1000 2026-01-03 Supplier_A
## 4 4 L4 100 2 1000 2026-01-04 Supplier_A
## 5 5 L5 100 1 1000 2026-01-05 Supplier_A
## 6 6 L6 100 2 1000 2026-01-06 Supplier_A
## 7 7 L7 100 4 1000 2026-01-07 Supplier_B
## 8 8 L8 100 5 1000 2026-01-08 Supplier_B
## 9 9 L9 100 6 1000 2026-01-09 Supplier_B
## 10 10 L10 100 4 1000 2026-01-10 Supplier_B
## 11 11 L11 100 7 1000 2026-01-11 Supplier_B
## 12 12 L12 100 8 1000 2026-01-12 Supplier_B
## observed_rate
## 1 0.01
## 2 0.00
## 3 0.01
## 4 0.02
## 5 0.01
## 6 0.02
## 7 0.04
## 8 0.05
## 9 0.06
## 10 0.04
## 11 0.07
## 12 0.08## n c producer_risk consumer_risk
## 1 132 3 0.04425251 0.0992283monitored <- dras_monitor(
prepared,
baseline_sample_size = plan$n,
acceptance_number = plan$c,
lambda = 0.40,
warning_threshold = 0.02,
tighten_factor = 2.00
)
monitored## lot lot_id sample_size sample_defects lot_size date supplier_id
## 1 1 L1 100 1 1000 2026-01-01 Supplier_A
## 2 2 L2 100 0 1000 2026-01-02 Supplier_A
## 3 3 L3 100 1 1000 2026-01-03 Supplier_A
## 4 4 L4 100 2 1000 2026-01-04 Supplier_A
## 5 5 L5 100 1 1000 2026-01-05 Supplier_A
## 6 6 L6 100 2 1000 2026-01-06 Supplier_A
## 7 7 L7 100 4 1000 2026-01-07 Supplier_B
## 8 8 L8 100 5 1000 2026-01-08 Supplier_B
## 9 9 L9 100 6 1000 2026-01-09 Supplier_B
## 10 10 L10 100 4 1000 2026-01-10 Supplier_B
## 11 11 L11 100 7 1000 2026-01-11 Supplier_B
## 12 12 L12 100 8 1000 2026-01-12 Supplier_B
## observed_rate ewma_rate inspection_mode recommended_next_sample_size
## 1 0.01 0.01000000 normal 132
## 2 0.00 0.00600000 normal 132
## 3 0.01 0.00760000 normal 132
## 4 0.02 0.01256000 normal 132
## 5 0.01 0.01153600 normal 132
## 6 0.02 0.01492160 normal 132
## 7 0.04 0.02495296 tightened 264
## 8 0.05 0.03497178 tightened 264
## 9 0.06 0.04498307 tightened 264
## 10 0.04 0.04298984 tightened 264
## 11 0.07 0.05379390 tightened 264
## 12 0.08 0.06427634 tightened 264
## lot_decision dras_signal
## 1 accept normal
## 2 accept normal
## 3 accept normal
## 4 accept normal
## 5 accept normal
## 6 accept normal
## 7 reject warning
## 8 reject warning
## 9 reject warning
## 10 reject warning
## 11 reject warning
## 12 reject warning## group n_lots accepted_lots rejected_lots normal_lots tightened_lots
## 1 All 12 6 6 6 6
## warning_lots mean_observed_rate max_observed_rate mean_sample_size
## 1 6 0.03416667 0.08 100
## total_sampled total_defects_found
## 1 1200 41## group n_lots accepted_lots rejected_lots normal_lots tightened_lots
## 1 Supplier_A 6 6 0 6 0
## 2 Supplier_B 6 0 6 0 6
## warning_lots mean_observed_rate max_observed_rate mean_sample_size
## 1 0 0.01166667 0.02 100
## 2 6 0.05666667 0.08 100
## total_sampled total_defects_found
## 1 600 7
## 2 600 34temp_dir <- tempfile("dras_report_")
files <- write_dras_report(monitored, output_dir = temp_dir)
files## [1] "/var/folders/j2/1c8pj_8j587ckrwklw3k4dnw0000gn/T//Rtmp3uw31a/dras_report_b55857e9a898/dras_lot_monitoring.csv"
## [2] "/var/folders/j2/1c8pj_8j587ckrwklw3k4dnw0000gn/T//Rtmp3uw31a/dras_report_b55857e9a898/dras_overall_summary.csv"
## [3] "/var/folders/j2/1c8pj_8j587ckrwklw3k4dnw0000gn/T//Rtmp3uw31a/dras_report_b55857e9a898/dras_supplier_summary.csv"The DRAS workflow provides an audit-oriented way to monitor whether observed defect rates are drifting upward. When the EWMA statistic exceeds the warning threshold, the system recommends tightened inspection for subsequent lots.