From StatsJam
[edit] Command Counts, No Command Names
On this page, we consider command usage in sessions, without regard to which commands.
See Per Command Statistics for summaries on a per-command basis.
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
COUNT(command_num) AS "Number of Commands"
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num
</query>
<view type="hist" caption="Histogram for Number of Commands Applied Per Session, Over All Log Files"/>
Histogram for Number of Commands Applied Per Session, Over All Log Files
In the histogram above, the number of commands applied is on the x-axis, while the number of sessions applying those commands is shown on the y-axis.
Because there appear to be fewer than 1000 commands applied per session, let's show scatter plots and histograms for sessions with fewer than 1000 commands:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
log_date AS "Log Date",
num_commands AS "Number of Commands"
FROM
interaction_log,
(SELECT
log_num,
COUNT(command_num) AS num_commands
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num
) AS command_count_table
WHERE
num_commands < 1000 AND
command_count_table.log_num = interaction_log.log_num
</query>
<view type="plot_date" caption="Scatterplot for Number of Commands Over All Log Files, For Log Files With Less Than 1000 commands"/>
Scatterplot for Number of Commands Over All Log Files, For Log Files With Less Than 1000 commands
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
num_commands
FROM
(SELECT
COUNT(command_num) AS num_commands
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num
) AS command_count_table
WHERE
num_commands < 1000
</query>
<view type="hist" caption="Histogram for Number of Commands Over All Log Files, For Log Files With Less Than 1000 commands"/>
Histogram for Number of Commands Over All Log Files, For Log Files With Less Than 1000 commands
Summarizing the data:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
ROUND(AVG(num_commands)) AS "Mean Number of Commands",
ROUND(MEDIAN(num_commands)) AS "Median Number of Commands",
ROUND(STDDEV(num_commands)) AS "SD Number of Commands",
COUNT(num_commands) AS "n"
FROM
(SELECT
log_num,
COUNT(command_num) AS num_commands
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num
) AS command_counts_table
</query>
<view type="table"/>
| Mean Number of Commands | Median Number of Commands | SD Number of Commands | n |
|---|
| 153.0 | 28.0 | 435.0 | 3917 |
Above is summary stats for number of commands applied per session.
Now on a per-user basis:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
ROUND(AVG(per_user_cmd_mean)) AS "Mean Mean Commands/User",
ROUND(MEDIAN(per_user_cmd_mean)) AS "Mean Median Commands/User",
ROUND(AVG(per_user_cmd_median)) AS "Median Mean Commands/User",
ROUND(MEDIAN(per_user_cmd_median)) AS "Median Median Commands/User",
COUNT(per_user_cmd_median) AS "n"
FROM
(SELECT
AVG(num_commands) AS per_user_cmd_mean,
MEDIAN(num_commands) AS per_user_cmd_median
FROM
interaction_log,
(SELECT
log_num,
COUNT(command_num) AS num_commands
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num
) AS command_counts_table
WHERE
interaction_log.log_num = command_counts_table.log_num
GROUP BY
user_id
) AS per_user_cmd_use_table
</query>
<view type="table" caption="Summary Stats for Number of Commands Per User"/>
Summary Stats for Number of Commands Per User| Mean Mean Commands/User | Mean Median Commands/User | Median Mean Commands/User | Median Median Commands/User | n |
|---|
| 116.0 | 56.0 | 63.0 | 26.0 | 212 |
At the time of this writing, the average user's average number of commands per session is 118; the median number of median commands by users is 26. So users are typically applying somewhere between 26-118 operations on images. Note that these numbers include undo/redo. We can filter those commands out:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
ROUND(AVG(per_user_cmd_mean)) AS "Mean Mean",
ROUND(MEDIAN(per_user_cmd_mean)) AS "Mean Median",
ROUND(AVG(per_user_cmd_median)) AS "Median Mean",
ROUND(MEDIAN(per_user_cmd_median)) AS "Median Median",
COUNT(per_user_cmd_mean) AS "n"
FROM
(SELECT
AVG(num_commands) AS per_user_cmd_mean,
MEDIAN(num_commands) AS per_user_cmd_median
FROM
interaction_log,
(SELECT
log_num,
COUNT(command_num) AS num_commands
FROM
native_command_events
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num
) AS command_counts_table
WHERE
interaction_log.log_num = command_counts_table.log_num
GROUP BY
user_id
) AS per_user_cmd_use_table
</query>
<view type="table" caption="Summary Stats for Number of Commands Per User"/>
Summary Stats for Number of Commands Per User| Mean Mean | Mean Median | Median Mean | Median Median | n |
|---|
| 92.0 | 36.0 | 52.0 | 15.0 | 211 |
At the time of this writing, this changes the average number of commands, per user, to 91, with a median median of 14 commands per session. Note that these numbers are not removing the commands that were undone.
[edit] Session Length Command Count Data Discussion
The number of commands per session again suggests that ingimp is being used for relatively small, targeted tasks, rather than extensive editing. Combining these data with the data above, users are applying about 1 command approximately every 9-21 seconds (if one removes undo/redo command counts). We turn now to data that considers the number of unique commands applied in each session.
[edit] Unique Commands in Sessions
Below are plots and histograms of the number of unique commands applied per session:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
log_date AS "Log Date",
num_unique_cmds AS "Number of Unique Commands"
FROM
interaction_log,
(SELECT
log_num,
COUNT(command_id) AS num_unique_cmds
FROM
(SELECT
DISTINCT event_record.log_num,
event_record.command_id
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num,
command_id
) AS unique_commands_per_log_table
GROUP BY
log_num
) AS unique_counts_log_table
WHERE
unique_counts_log_table.log_num = interaction_log.log_num
</query>
<view type="plot_date" caption="Scatter Plot for Number of Unique Commands Per Log File, Over All Log Files"/>
Scatter Plot for Number of Unique Commands Per Log File, Over All Log Files
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
log_date AS "Log Date",
num_unique_cmds AS "Number of Unique Commands"
FROM
interaction_log,
(SELECT
log_num,
COUNT(command_id) AS num_unique_cmds
FROM
(SELECT
DISTINCT event_record.log_num,
event_record.command_id
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num,
command_id
) AS unique_commands_per_log_table
GROUP BY
log_num
) AS unique_counts_log_table
WHERE
unique_counts_log_table.log_num = interaction_log.log_num AND
num_unique_cmds < 40
</query>
<view type="plot_date" caption="Scatter Plot for Number of Unique Commands Per Log File With Fewer Than 40 Unique Commands, Over All Log Files"/>
Scatter Plot for Number of Unique Commands Per Log File With Fewer Than 40 Unique Commands, Over All Log Files
With the data summarized:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
ROUND(AVG(num_unique_cmds)) AS "Mean",
ROUND(MEDIAN(num_unique_cmds)) AS "Median",
ROUND(STDDEV(num_unique_cmds)) AS "SD",
MAX(num_unique_cmds) AS "Max",
MIN(num_unique_cmds) AS "Min",
COUNT(num_unique_cmds) AS "n"
FROM
(SELECT
log_num,
COUNT(command_id) AS num_unique_cmds
FROM
(SELECT
DISTINCT event_record.log_num,
event_record.command_id
FROM
event_record
WHERE
log_num IN (SELECT log_num FROM significant_logs)
GROUP BY
log_num,
command_id
) AS unique_commands_per_log_table
GROUP BY
log_num
) AS unique_cmd_count_table
</query>
<view type="table"/>
| Mean | Median | SD | Max | Min | n |
|---|
| 11.0 | 8.0 | 10.0 | 91 | 1 | 3917 |
Summary Stats for the Number of Unique Commands Per Log File, Over All Log Files
Now the data looked at per user:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
COUNT(command_id) AS " "
FROM
(SELECT
DISTINCT interaction_log.user_id,
event_record.command_id
FROM
event_record,
interaction_log
WHERE
interaction_log.user_id IN (SELECT user_id FROM significant_users) AND
interaction_log.log_num = event_record.log_num
GROUP BY
interaction_log.user_id,
command_id
) AS unique_commands_per_user_table
GROUP BY
user_id
</query>
<view type="hist" caption="Histogram for Number of Unique Commands Per User, Over All Sessions for the User"/>
Histogram for Number of Unique Commands Per User, Over All Sessions for the User
Summarized:
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
ROUND(AVG(num_unique_cmds)) AS "Mean",
ROUND(MEDIAN(num_unique_cmds)) AS "Median",
ROUND(STDDEV(num_unique_cmds)) AS "SD",
MAX(num_unique_cmds) AS "Max",
MIN(num_unique_cmds) AS "Min",
COUNT(num_unique_cmds) AS "n"
FROM
(SELECT
user_id,
COUNT(command_id) AS num_unique_cmds
FROM
(SELECT
DISTINCT interaction_log.user_id,
event_record.command_id
FROM
event_record,
interaction_log
WHERE
interaction_log.user_id IN (SELECT user_id FROM significant_users) AND
interaction_log.log_num = event_record.log_num
GROUP BY
interaction_log.user_id,
command_id
) AS unique_commands_per_user_table
GROUP BY
user_id
) AS unique_cmd_count_table
</query>
<view type="table" caption="Summary Stats for Number of Unique Commands Per User, Over All Sessions for the User"/>
Summary Stats for Number of Unique Commands Per User, Over All Sessions for the User| Mean | Median | SD | Max | Min | n |
|---|
| 43.0 | 34.0 | 30.0 | 189 | 4 | 212 |
From the above data, users have employed (or tried out) an average of 41 different, unique commands (with 32 commands the median)
over all recorded sessions. Now let's consider the typical number of unique commands they employ in a
single session.
This form gives you a chance to modify the query and quickly see
the results but it does not update the wiki page..
<query>
SELECT
ROUND(AVG(mean_per_user)) AS "Mean Mean Per User",
ROUND(MEDIAN(mean_per_user)) AS "Median Mean Per User",
ROUND(STDDEV(mean_per_user)) AS "SD Mean Per User",
ROUND(AVG(median_per_user)) AS "Mean Median Per User",
ROUND(MEDIAN(median_per_user)) AS "Median Median Per User",
ROUND(STDDEV(median_per_user)) AS "SD Median Per User",
COUNT(mean_per_user) AS "n"
FROM
(SELECT
AVG(num_unique_cmds) AS mean_per_user,
MEDIAN(num_unique_cmds) AS median_per_user
FROM
(SELECT
user_id,
COUNT(command_id) AS num_unique_cmds
FROM
(SELECT
DISTINCT event_record.log_num,
interaction_log.user_id,
event_record.command_id
FROM
event_record,
interaction_log
WHERE
interaction_log.user_id IN (SELECT user_id FROM significant_users) AND
interaction_log.log_num = event_record.log_num
) AS unique_commands_per_user_log_table
GROUP BY
log_num,
user_id
) AS unique_cmd_count_table
GROUP BY
user_id
) AS user_summary_table
</query>
<view type="table" caption="Summary Stats for Number of Unique Commands Per User Per Session"/>
Summary Stats for Number of Unique Commands Per User Per Session| Mean Mean Per User | Median Mean Per User | SD Mean Per User | Mean Median Per User | Median Median Per User | SD Median Per User | n |
|---|
| 11.0 | 10.0 | 5.0 | 9.0 | 8.0 | 5.0 | 212 |
From these data, we see the numbers closely follow the summary for unique commands on a per-log file basis: The mean mean, per user, is 11 unique commands per session, with the median median number of commands per session 8.
[edit] Unique Command Discussion
The data above indicate that users are again performing rather targeted tasks, employing only a dozen or fewer tools to accomplish their tasks in each ingimp session.
