-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathnotes.txt
192 lines (153 loc) · 8.44 KB
/
notes.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
Some Mechanisms Related to Interest Development
AGENTS
Child/Youth
Each begins with a level of interest in a particular cluster of science topics (e.g., ÒastronomyÓ).
When a child/youth has an interaction in a science activity, their interest level changes.
Peers
When friends are participants in an activity, the value to interest increases (below the interest threshold).
When no friends are participants in an activity, the value to interest decreases (below the interest threshold).
Above the interest threshold, children/youth have a certain probability of making friends with others in the activity.
Parents/Caregivers
When parents are participants in an activity, the value to interest increases.
When no parents are participants in an activity, the value to interest decreases.
When parents actively broker access to activities to allow children to pursue interest, value to interest increases.
Unrelated Adults
Unrelated adults are associated with interactions in activities; each has an arbitrary level of expertise and passion in the topic of that activity.
There are some arbitrary fixed number of unrelated adults available for interactions in the community.
When unrelated adults are skilled at sharing their expertise and passion in a topic in an activity, the value to interest increases.
When unrelated adults are skilled at sharing expertise but not passion in a topic in an activity, the value to interest stays the same (unless the interest threshold has been exceeded, then it increases).
When unrelated adults are skilled at sharing passion but not expertise in a topic in an activity, the value to interest stays the same.
When unrelated adults are neither skilled nor passionate, the value to interest decreases.
Web resource
Web resources vary in their fit to interest and expertise level of youth.
Web resources that are a good fit to both increase interest.
Web resources that are a poor fit to both decrease interest.
Relative to interactions with people,
INTERACTIONS
Interactions among agents take place within science activities.
Activities can take place in physical spaces or online.
Interactions include the child/youth, and they can also include parents, peers, and unrelated adults. Or, interactions can be with the self, interacting with a web-based resource.
There are school interactions and non-school interactions.
School interactions with unrelated adults in science activity are a weekly occurrence.
Up to a certain age (say 10-11), interactions in non-school science activity depend on access determined by parents.
Beyond age 10-11 but below a certain threshold of interest, non-school interactions are random, and dependent on available activities in the community.
Beyond age 10-11 and above a certain threshold of interest, non-school interactions can be pursued by the self (occur more than chance), but are still dependent on available activities in the community.
ENVIRONMENTS
Children/youth live in a particular place and have a spatial zone of weekly activity within which most of their activities occur.
The environment has a fixed number of potential science activities that are located in particular places.
School is a place where children have regular (weekly) interactions with science activities.
Occasionally, children/youth venture out of this zone to pursue science-related activities (e.g., zoo, science museum).
The spatial distance of non-school science activities from the child matter: the farther away from home, the less likely to pursue (all other things being equal).
Rules for making new friends:
During an activity,
0) if no friends are present,
if random < makeFriendProb,
pick a new friend with highly similar interests
1) if one friend is present,
if friend of friend is present,
if random < closeTriadProb,
close the triad
else
same as 0)
2) if two or more friends are present,
do nothing
-------
Info about Prescott Elementary
First day of schoolyear: Tuesday, Sept 4th 2012 (jday: 247)
Winter break Dec 24 - Jan 4
Last day of schoolyear: Friday, June 7th 2013 (jday: 158)
In 2013:
199 school days (ignoring holidays/breaks)
104 weekend days
78 during the school year
26 during summer
88 summer days
62 during the week
26 during the weekend
-------
TODO:
Data logging:
[ ] Effect of individual activities
Graphs:
[x] Histogram of interest distribution (x3)
[x] Histogram of number of activities done per day
[x] Histogram of how many times each activity is being done (minus class maybe)
Ideas from meeting (12/16):
[ ] Change the effect of parent participation as kids get older
[ ] Heterogeneous student parameter for interest change rate
[ ] Unrelated adults might have an emotional intelligence parameter (not sure what this means)
[ ] Disruption caused by the changing of leaders
[x] Vary level of adult expertise and passion
[ ] Come up with some way of initializing expertise and passion that isn't just random
[ ] GIS
[ ] Families
[ ] Schools
[ ] Resources (e.g. museums, etc.)
[ ] Make the Reset button work in the network display
[ ] Look into the literature on the effect of small world rewire probability
[ ] Init model parameters from properties to allow sweeps
[x] Make NetworkDisplay hide on close
[ ] Do multiple runs with random interests and stuffIDo to get average and error bars (make script)
[ ] Add a mechanism for consensus formation ("balance seeking")
[x] Use daysBetween (or something) to reduce frequency of activities
[/] Use Nancy's correlation data to populate activity topic vectors
[x] Randomize order of activities when scheduling to eliminate bias
[ ] Find a way to quantify parent involvement and use it to determine if parent is an activity leader
[ ] Add column(s) to student data file to represent parent encouragement level
[ ] Add a column to activity type to specify activies as dynamically tailored to individual interest
[x] Switch the model over to use the new Rule types
[x] Use different node shapes for male (square) and female (round)
[x] Make the organized activities (scouts, after-school, summer camp) repeat with the same participants
[x] Add a column to activityTypes.csv for scheduling priority (i.e. school and other activities are higher than ad hoc ones)
[ ] Create a new spreadsheet for adults who'll lead different activities
[x] Design an algorithm for matching students to activity instances (of a given type) that keep friends together
[x] Change everything in the model to run for 4 years instead of 3. (including csv files)
[x] Bring MersenneTwister into the project and use it for shuffling.
[ ] MakeFriendRule needs to consider inter-gender friend probability
[x] Verify that the correct number of repeating activities are being created
[x] Create a mechanism to allow for schedule adjustment (synergy)
Three scenarios: no coordination, some coordination, lots of coordination
[x] Create a chart that shows the cumulative effect of each activity
[x] Create a chart that shows the cumulative effect of each rule
[ ] isSchoolRelated isn't used. Should it be?
[x] Add adult leaders to organized activities
[x] Rename UnrelatedAdultRule to LeaderRule
[x] Sort Students by 5th grade teacher before grouping them into ScienceClasses.
[ ] Make sure priority sorting is working properly (0 is highest priority)
Individual-level visualization
[ ] Interest vectors
[ ] Activities
[ ] Friend network
Idea for visualization:
GIS of area showing house locations for each student, draw links connecting friends.
Links will change over time to be based less on geography and more on interests.
[x] Histogram for each activity showing number of times each student has done that activity
See:
Add Health Survey for a dataset of adolescents
r = i + epsilon
epsilon = N(0,sigma^2)
Issues with the rules:
Solo activities always result in a decrease in interest
[x] Make rules only have effect when they are relevant, e.g. with solo activities, lack of friends isn't relevant
Problems with data:
- IDs
- School
- Teacher
Potential Experiments:
- Would higher teacher expertise yield different results?
-----------------------
Evolving the mapping from stuffIDo responses to participation rates. Initially, we used
a naive mapping where:
Never: 0
Very Rarely: 0.25
Every once in a while: 0.5
Often: 0.75
All the time: 1.0
In the second round of surveys, they changed the wording of the question to be more precise.
Thus, the mapping becomes:
Never: 0
Few times a year: 0.25
1-2 a month: 0.5
1-2 a week: 0.75
Almost every day: 1.0
These values are almost certainly wrong, and should probably be different for each activity.