If you use Anki for spaced repetition learning, you’ve probably wondered about your study patterns. How many cards have you reviewed? What’s your retention like? Which cards are giving you trouble?
I built ankiR to make this easy in R.
The Problem
Anki stores everything in a SQLite database, but accessing it requires writing raw SQL queries. Python users have ankipandas, but R users had nothing—until now.
Installation
# From GitHub
remotes::install_github("chrislongros/ankiR")
# Arch Linux (AUR)
yay -S r-ankir
Basic Usage
ankiR auto-detects your Anki profile and provides a tidy interface:
library(ankiR)
# See available profiles
anki_profiles()
# Load your data as tibbles
notes <- anki_notes()
cards <- anki_cards()
reviews <- anki_revlog()
# Quick stats
nrow(notes) # Total notes
nrow(cards) # Total cards
nrow(reviews) # Total reviews
FSRS Support
The killer feature: ankiR extracts FSRS parameters directly from your collection.
FSRS (Free Spaced Repetition Scheduler) is the modern scheduling algorithm in Anki that calculates optimal review intervals based on your memory patterns.
fsrs_cards <- anki_cards_fsrs()
This gives you:
- stability – memory stability in days (how long until you forget)
- difficulty – card difficulty on a 1-10 scale
- retention – your target retention rate (typically 0.9 = 90%)
- decay – the decay parameter used in calculations
Example: Visualize Your Card Difficulty
library(ankiR)
library(dplyr)
library(ggplot2)
anki_cards_fsrs() |>
filter(!is.na(difficulty)) |>
ggplot(aes(difficulty)) +
geom_histogram(bins = 20, fill = "steelblue") +
labs(
title = "Card Difficulty Distribution",
x = "Difficulty (1-10)",
y = "Count"
) +
theme_minimal()
Example: Stability vs Difficulty
anki_cards_fsrs() |>
filter(!is.na(stability)) |>
ggplot(aes(difficulty, stability)) +
geom_point(alpha = 0.3, color = "steelblue") +
scale_y_log10() +
labs(
title = "Memory Stability vs Card Difficulty",
x = "Difficulty",
y = "Stability (days, log scale)"
) +
theme_minimal()
Example: Review History Over Time
anki_revlog() |>
count(review_date) |>
ggplot(aes(review_date, n)) +
geom_line(color = "steelblue") +
geom_smooth(method = "loess", se = FALSE, color = "red") +
labs(
title = "Daily Review History",
x = "Date",
y = "Reviews"
) +
theme_minimal()
Calculate Retrievability
You can also calculate the probability of recalling a card after N days:
# What's my retention after 7 days for a card with 30-day stability?
fsrs_retrievability(stability = 30, days_since_review = 7)
# Returns ~0.93 (93% chance of recall)
Links
I built this because I wanted to analyze my USMLE study data in R. If you find it useful, let me know!
/root 