Last updated: 2021-02-19
Checks: 7 0
Knit directory:
fa_sim_cal/
This reproducible R Markdown analysis was created with workflowr (version 1.6.2). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.
Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.
Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.
The command set.seed(20201104) was run prior to running the code in the R Markdown file.
Setting a seed ensures that any results that rely on randomness, e.g.
subsampling or permutations, are reproducible.
Great job! Recording the operating system, R version, and package versions is critical for reproducibility.
Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.
Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.
Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.
The results in this page were generated with repository version 815af1e. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.
Note that you need to be careful to ensure that all relevant files for the
analysis have been committed to Git prior to generating the results (you can
use wflow_publish or wflow_git_commit). workflowr only
checks the R Markdown file, but you know if there are other scripts or data
files that it depends on. Below is the status of the Git repository when the
results were generated:
Ignored files:
Ignored: .Rhistory
Ignored: .Rproj.user/
Ignored: .tresorit/
Ignored: data/VR_20051125.txt.xz
Ignored: output/blk_char.fst
Ignored: output/ent_blk.fst
Ignored: output/ent_cln.fst
Ignored: output/ent_raw.fst
Ignored: renv/library/
Ignored: renv/staging/
Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.
These are the previous versions of the repository in which changes were made
to the R Markdown (analysis/notes.Rmd) and HTML (docs/notes.html)
files. If you’ve configured a remote Git repository (see
?wflow_git_remote), click on the hyperlinks in the table below to
view the files as they were in that past version.
| File | Version | Author | Date | Message |
|---|---|---|---|---|
| Rmd | 815af1e | Ross Gayler | 2021-02-19 | Add sampling to notes |
| Rmd | a2cde2d | Ross Gayler | 2021-02-17 | end of day |
| Rmd | a093655 | Ross Gayler | 2021-02-16 | end of day |
| Rmd | 0935c13 | Ross Gayler | 2021-02-10 | end of day |
| html | 0d30c5b | Ross Gayler | 2021-01-26 | Build site. |
| html | 23d740f | Ross Gayler | 2021-01-24 | Build site. |
| Rmd | d0307f5 | Ross Gayler | 2021-01-24 | Add 02-1 char block vars |
| Rmd | 6df8db7 | Ross Gayler | 2021-01-24 | End of day |
| Rmd | 3090bea | Ross Gayler | 2021-01-23 | End of day |
| Rmd | 94b7b20 | Ross Gayler | 2021-01-19 | End of day |
| html | 0c48ca5 | Ross Gayler | 2021-01-17 | Build site. |
| Rmd | 3052780 | Ross Gayler | 2021-01-17 | Add 02-1 block vars |
| html | 3eb6e3b | Ross Gayler | 2021-01-16 | Build site. |
| Rmd | e6721d3 | Ross Gayler | 2021-01-16 | Add blocking to notes.Rmd |
| html | 5ab5dc4 | Ross Gayler | 2021-01-15 | Build site. |
| Rmd | c674a51 | Ross Gayler | 2021-01-15 | Add 01-6 clean vars |
| html | 0b67848 | Ross Gayler | 2021-01-06 | Build site. |
| Rmd | ec38ffc | Ross Gayler | 2021-01-06 | wflow_publish("analysis/no*.Rmd") |
| html | 03ad324 | Ross Gayler | 2021-01-05 | Build site. |
| Rmd | b03a2c1 | Ross Gayler | 2021-01-05 | wflow_publish(c(“analysis/index.Rmd”, “analysis/notes.Rmd”)) |
This document is for keeping notes of any points that may be useful for later project or manuscript development and which are not covered in the analysis notebooks or at risk of getting lost in the notebooks.
Get a sizeable publicly available data set with personal names (NCVR).
Use sex and age in addition to personal names so that most records are discriminable.
High frequency names will likely not be discriminable with only these attributes.
Age (and possibly sex) will be used as a blocking variable.
Age and sex are also of interest in the calculation of name frequency because name distributions should vary conditional on age and sex.
Keep address and phone number as they may be useful for manually checking identity in otherwise nondiscriminable records.
Get the oldest available data to minimise it’s currency (NCVR 2005 snapshot).
Drop objectionable attributes such as race and political affiliation.
A note about the irregularity of names: https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/
Apply basic data cleaning to the predictive attributes.
This is probably unnecessary given how the data will be used.
I can’t bring myself to model data without scrutinising it first.
Only keep records that are ACTIVE and VERIFIED for modelling.
These are likely to have the highest data quality attributes.
These are least likely to have duplicate records (i.e. referring to the same person).
Use blocking to reduce the number of comparisons, to keep the computational cost feasible.
This project is not an operational system and we are only using blocking to reduce the computational cost, so we can choose blocking that would not be acceptable for an operational system.
Where the dictionary blocks vary widely in size we might choose to work with only a subset of blocks that are a suitable size.
If we think that some aspect of the compatibility modelling might vary as a function of block size we would probably want to test this over a wide range of block sizes.
We will probably be repeating analyses over some number of blocks to assess the variability of results, but that might only be a subset of blocks with no commitment to examine all the blocks.
Blocking variables may have missing values in the query and dictionary records.
Exclude records with missing values on the blocking records from use, so that missing values don’t have to be allowed for in blocking.
Try to choose blocking variables with a small proportion of missing values, so as to minimise systematic bias due to their exclusion.
Construct a few potentially useful blocking variables.
Blocking can induce changes in the distributions of names.
Blocking on sex (in combination with other variables) will definitely give more homogeneity of first names within blocks because of gendered names.
Blocking on age will give more homogeneity of first names within blocks because of name popularity varying over time.
Blocking on county may give more homogeneity of last names within blocks because of families living together.
These sections are about looking at the properties of the entity data that will be most relevant to their use in the compatibility models.
Prior work on name frequency has (implicitly) only considered equality of names. That is two name tokens are either absolutely identical or absolutely different. (No gradations of difference are entertained.) Counting the frequency of names is finding the cardinality of each set of identical name tokens.
Look at frequency distributions of names conditional on name length. The Zipf distributions may have different shape parameters for different name lengths. Name length might be examined as an alternative to name frequency for interaction with similarity.
Look at frequency distributions of names conditional on age and/or sex.
Look at frequency distributions of names conditional on blocking variables.
The similarity version of name frequency is an extension of the equality version. It counts the number of dictionary records in the block that are at least as similar to the query record as the currently considered dictionary record.
Look at similarity frequency distributions of names. It’s not obvious that these should be Zipf distributions. For example, the rare names might be quite similar to more frequent names, which might obscure the long tail of the underlying distribution of names.
Look at similarity frequency distributions of names conditional on name length. The Zipf distributions may have different shape parameters for different name lengths.
Look at similarity frequency distributions of names conditional on age and/or sex.
Look at similarity frequency distributions of names conditional on blocking variables.
Both modelling and performance evaluation will be based on random samples of the entity data, so sort out the sampling first.
Modelling is necessarily based on query/dictionary record pairs. That is, the record pair is the unit of analysis.
Performance assessment of model calibration will also be at the record pair level.
Performance assessment of overall model performance at identifying the correctly matching dictionary record (e.g. AUC or F-score) is at the query record level.
The query/dictionary record pairs are generated as combinations of a set of queries with a fixed dictionary. The record pairs can’t be uniformly sampled from the set of all possible record pairs because that wouldn’t respect the dictionary definition.
We need some sort of training/testing partition to keep performance assessment honest.
Remember that queries are run against a dictionary, so in principle, it would be possible to have separate test/train queries run against one fixed dictionary.
I am inclined to think that running training/testing queries against the same dictionary is probably OK because in a practical application the modelling would be done with respect to the dictionary as it existed at some point in time and then queries applied against essentially the same dictionary. (Given a large dictionary, subsequent queries would update it relatively slowly.)
On the other hand, I am inclined to think that using separate dictionaries for training/test would not cause a problem, because the proposed models are not intended to rely on the actual names in the dictionary, but on properties of the names (like frequency and length), which are likely to be very similar for different dictionaries based on random samples of the universe of names.
On balance, I think I will have different dictionaries for train/test because it is probably harmless, probably conservative with respect to performance assessment, and probably easy to implement.
Dictionaries are constructed from entity data records, so having separate train/test dictionaries implies partitioning the entity data records prior to constructing the dictionaries.
The easiest way to partition the records is to partition independently at the record level.
Partitioning the entity records so that each name type occurred in only one of the train or test data sets would be the most conservative in terms of separation of the train and test data sets.
Partition the entity data records uniformly and independently into training and testing data sets.
We will want to repeat the entire analytic process multiple times on different samples in order to assess the variability of the results.
Repeat the partitioning into training and test sets for as many replicates as are needed.
The number of replicates is yet to be determined.
It needs to be large enough to give a reasonable view of variability of the results.
It needs to be not so large as to be too computationally expensive.
The dictionary will be a randomly sampled subset of the test or train universe, because it is usual in reality for the dictionary to be a subset of the universe of entities.
The dictionary sampling will be repeated at a range of sampling rates. This will allow us to investigate calibration of the model as the fraction of the universe present in the dictionary varies.
It is expected that this can be captured as a change of value of the intercept term of the logistic regression.
In practice, we usually don’t know the size of the universe relative to the dictionary.
If this effect is captured as an intercept term it will be simple to adjust the model to reflect the assumed size of the universe.
The queries are selected as a random subset of the relevant universe.
Use the same queries across all the corresponding dictionaries.
This is marginally less effort than choosing different queries per dictionary
This should result in the evaluations across dictionaries being somewhat more similar than if the queries were chosen separately for each dictionary, because the “difficulty” of the queries is fixed across dictionaries.
The fraction of queries having matches in the dictionary will (obviously) vary with the sampling rate that created the dictionary.
Sample some fixed number of queries (yet to be determined)
The number of queries needs to be large enough to reasonably cover the blocks in the directory, the range of the predictors used in the model, and provide a reasonable observations to parameters ratio for estimating the model.
The number of queries needs to be not so large as to be infeasibly expensive to compute the models.
The entire sampling process is summarised in the following diagram.
The square nodes represent sets of entity data records.
The triangular nodes represent random sampling processes.
The text-only nodes represent replicates of random sampling nodes (just made less visually obtrusive, for clarity).
The top data node represents the complete data set of entity data records.
The “train/test 50:50 splitter 1” node randomly partitions the total data into two data sets of (approximately) equal size: the “train universe” and the “test universe”.
The “train/test 50:50 splitter” nodes represent the replication of the train/test split as many times as are needed.
The “query sampler: n_q” node randomly samples \(n_q\) records from the “train universe” to be used as the “train queries” to be run against all of the dictionaries sampled from the “train universe”.
The “dictionary sampler: 90%” node randomly selects 90% of the “train universe” records to use as a dictionary.
The remaining “dictionary sampler” nodes create more dictionaries with different selection rates.
The one query data set is applied to all the corresponding dictionary data sets.
The construction of the query and dictionary data sets from the “test universe” exactly parallels the construction from the “train universe”.
The construction of the data sets under each of the “train/test 50:50 splitter” nodes exactly parallels the construction under the “train/test 50:50 splitter 1” node.
Try indicators for missingness. Missingness may be differentially informative across different predictor variables.
Try indicators for similarity == 1. The compatibility of exact string equality is not necessarily continuous with the compatibility of similarity just below 1.
Try name frequency as an interactive predictor variable.
There are two names in each lookup: dictionary and query. Therefore there are also two name frequencies to be considered.
We are interested in calculating the probability of a correct match for each query, so the sought probability is conditional on the query. Consequently the conditional probability of a correct match should not depend on the frequency of the query name.
I am not entirely convinced by that argument, so consider how to use both frequencies (e.g. min, max, geometric mean, …).
Queries may contain names that do not exist in the dictionary, so we need to deal with that case.
Do we need to apply frequency smoothing, as used in probabilistic linguistic models?
Do we need to estimate the probability mass of unobserved names?
In general, the dictionary will be a subset of the entities in the universe of queries. Consider the impact of this on modelling as the fraction of the query universe in the dictionary varies.
Can the sampling fraction be modelled as a prior probability (effectively, a change in the intercept term of a logistic regression)?
This can be investigated by varying the fraction of dictionary records randomly sampled from the universe and treating the sampling fraction as a predictor in the model. The fitted coefficients for the sampling fraction can be compared to what would be expected treating the same sampling probabilities as a prior for the logistic regression.
Consider whether this varies on a per query basis because of blocking. That is, is there effectively a separate dictionary per blocking value?
Consider using the same variable for blocking and as a predictor to compare the effect on estimated probability of identity match as a function of dictionary fraction.
Performance evaluation should always be performed on a disjoint set of records from those used in estimation of the model.
The obvious (and probably best) way to partition the data is to randomly select records. This means that for more frequent names, the same name will appear in the fitting and testing sets. So, if the model is in some sense “learning” details of individual names, this knowledge will be carried over from the fitting to testing sets and the performance of the model will be overstated.
An alternative approach would be to partition the name types (as opposed to tokens). This would ensure that the names in the fitting and testing sets were completely disjoint, completely excluding any possibility of name-specific information being carried from fitting to testing.
Partition the entity data records into a dictionary and a set of queries (the \(Q_U\) set).
Select a subset of the dictionary records to use as the \(Q_M\) query set.
This is evaluation is different to the usual evaluation of entity resolution in that it doesn’t consider the impact of transcription/typographical variation in the queries.
If we are interested in the performance with respect to transcription/typographical variation we may need to consider artificially corrupting some of the queries.
Consider assigning some of the dictionary records to randomly chosen wrong blocks. (Is this equivalent to randomly selecting some out-of-block query records to run against each dictionary block?)
The unit of analysis is a dictionary/query record pair.
The logistic regression model is modelling the probability that the dictionary record refers to the same entity as the query record.
That is, the record-pair modelling yields an unconditional probability that does not depend on how that predicted probability will be used (e.g. compared to a threshold).
However, the logic behind the name frequency induced by similarity does appear to incorporate a dependence on usage. The frequency is calculated as the sum of the frequncies of the names (in the same block) that have similarity to the query name greater than or equal to the query similarity to the dictionary name being considered. This is based on an implicit ceteris paribus usage of similarity: If I were to accept this particular name as indicating an entity match, then logically I should also accept all other names that are at least as similar.
The name frequency induced as equality can be construed as compatible with this definition by interpreting equality as a degenerate similarity relation.
Is this a reasonable argument?
Should this implied comparison across names within block be incorporated somehow in the mathematical formulae?
It might be possible to develop (or at least explain) most of the maths in terms of blocking.
Start with a universe \(E\) of entities with the queries drawn uniformly at random from that universe and the dictionary being identical to the universe.
In the absence of any other information, the probability of each dictionary element being an identity match with the query is $ 1 / |E| $.
If the dictionary \(D\) is a proper subset of \(E\) it can be thought of as a (not very informative) block. That is, a block can be thought of as the dictionary induced by the query record.
If \(D\) is a random subset of \(E\), and in the absence of any other information, the probability of each dictionary element being an identity match with the query is \((|D| / |E|) / |D| = 1 / |E|\).
\(|D|\) is the block size of the dictionary construed as a block. In the absence of any information to discriminate between records in the block, the larger the block, the lower the probability the any block record is the identity match to the query record.
\(|D| / |E|\) is the probability that the entity corresponding to the query record is in the dictionary/block, given that the dictionary is a uniform random subset of the universe of entities.
Consider the other extreme, where blocking is perfect. That is, the entity query record is guaranteed to be in the dictionary. In this case the blocking is very informative (not a random selection from the universe of entities). The probability that the correctly matching entity record is in the block is 1, and the probability of each dictionary/block element being an identity match with the query is \(1 / |D|\).
That is, with perfect blocking, the probability of each entity being the identity match depends only on the block size.
Where blocking is based on, say, name, the block size is the frequency of the name in the dictionary.
Where blocking is less then perfect the probability that the entity corresponding to the query record is contained in the block will be less than 1.
This suggests that the probability of each record in the block being the correct match could be calculated as the product of conditional probabilities:
\[ P(id(q) = id(d_i)) = P(id(q) = id(d_i) | d_i \in B_j) P(d_i \in B_j | d_i \in D) P(d_i \in D) \]
What we need are models to estimate those component probabilities. I suspect that the cardinalities of those blocks would be very strong predictors of the probabilities. If the blocks are defined in terms of equality of names then the name frequencies determine the block cardinalities.
If there were multiple independent blockings, the probability of correct match of each record in the intersection of the blocks id the product of the probabilities associated with each block. This is equivalent to naive Bayes and shows the equivalence between construing the problem as a multivariate regression and construing it as using multiple blocking variables.
R version 4.0.3 (2020-10-10)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 20.10
Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.9.0
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.9.0
locale:
[1] LC_CTYPE=en_AU.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_AU.UTF-8 LC_COLLATE=en_AU.UTF-8
[5] LC_MONETARY=en_AU.UTF-8 LC_MESSAGES=en_AU.UTF-8
[7] LC_PAPER=en_AU.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_AU.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices datasets utils methods base
other attached packages:
[1] DiagrammeR_1.0.6.1 here_1.0.1 workflowr_1.6.2
loaded via a namespace (and not attached):
[1] Rcpp_1.0.6 rstudioapi_0.13 whisker_0.4 knitr_1.31
[5] magrittr_2.0.1 R6_2.5.0 rlang_0.4.10 stringr_1.4.0
[9] visNetwork_2.0.9 tools_4.0.3 xfun_0.21 git2r_0.28.0
[13] htmltools_0.5.1.1 ellipsis_0.3.1 yaml_2.2.1 digest_0.6.27
[17] rprojroot_2.0.2 tibble_3.0.6 lifecycle_1.0.0 bookdown_0.21
[21] crayon_1.4.1 RColorBrewer_1.1-2 later_1.1.0.1 htmlwidgets_1.5.3
[25] vctrs_0.3.6 promises_1.2.0.1 fs_1.5.0 glue_1.4.2
[29] evaluate_0.14 rmarkdown_2.6 stringi_1.5.3 compiler_4.0.3
[33] pillar_1.4.7 jsonlite_1.7.2 httpuv_1.5.5 renv_0.12.5
[37] pkgconfig_2.0.3