[{"data":1,"prerenderedAt":456},["ShallowReactive",2],{"project-31238396":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":47,"post_text":48,"what":49,"said":93,"maker_in_thread":227,"shape":229,"grid":233,"attempts":414,"same_week":419,"maker_shows":46,"maker_hidden":448,"reader_version":449,"card":450},"31238396","PostgresML","We've been hard at work for a few weeks and thought it's time for another update.","post","Show HN: PostgresML, now with analytics and project management","https:\u002F\u002Fpostgresml.org","postgresml.org","levkk",1651513718,2022,444,70,72,2570,3,"legend","ai","ml in database","library",[24,25,26,27],"postgres machine learning","in-database ml","postgresql extension","sql machine learning","curious",0.108,{"positive":31,"neutral":32,"mixed":33,"critical":34},15,33,1,10,59,{"idea":37,"execution":40,"performance":41,"docs":43,"design":45},{"positive":38,"critical":39},9,4,{"positive":33,"critical":33},{"positive":42,"critical":42},0,{"positive":44,"critical":42},2,{"positive":33,"critical":46},5,[46,46,33,42,46,46,18,46,18,18,18,33,18,18,46,18,18,18,18,42,33,18,46,46,33,18,46,46,18,18,18,46,33,46,18,18,18,33,18,33],"We've been hard at work for a few weeks and thought it's time for another update.\n\nIn case you missed our first post, PostgresML is an end-to-end machine learning solution, running alongside your favorite database.\n\nThis time we have more of a suite offering: project management, visibility into the datasets and the deployment pipeline decision making.\n\nLet us know what you think!\n\nDemo link is on the page, and also here: https:\u002F\u002Fdemo.postgresml.org",{"problem":50,"audience":62,"pricing":69,"motivation":69,"time_to_build":69,"stack":70,"openness":91,"stage":92,"team_size":91},{"value":51,"source":52,"evidence":53},"Running ML on data without exporting it from the database into separate Python pipelines","commenter",[54,58],{"id":55,"by":56,"text":57},"31241685","zmmmmm","Seems like a great idea. When you look at many ML frameworks half the code and learning overhead is data schlepping code and table like structures that \"reinvent\" the schema that…",{"id":59,"by":60,"text":61},"31240529","montanalow","OLAP vs OLTP will depend on your ML use case. Online predictions will likely be better served by an OLTP vs offline batch predictions being better served by OLAP. OLAP use cases…",{"value":63,"source":52,"evidence":64},"Businesses and startups with data in SQL databases who want simple ML models",[65],{"id":66,"by":67,"text":68},"31240723","jdoliner","This is really cool, running ML workloads on top of SQL is a very practical way of doing ML for a lot of businesses. Many companies don't have the fancy ML workloads like you see…",null,[71,76,87],{"value":72,"source":73,"evidence":74},"PostgreSQL","maker",[75],{"id":7},{"value":77,"source":52,"evidence":78},"Python",[79,83],{"id":80,"by":81,"text":82},"31244077","mr_toad","It looks like this is running Python to do the actual ML? There’s nothing stopping you from reading database table structures directly into memory in Python or R now. You don’t…",{"id":84,"by":85,"text":86},"31242220","ekzhu","Great idea! I see this is implemented using the Python language interface supported by PostgreSQL and importing sklearn models. I always wonder how scalable this is considering…",{"value":88,"source":52,"evidence":89},"scikit-learn",[90],{"id":84,"by":85,"text":86},"unknown","beta",{"praise":94,"objections":110,"requests":141,"questions":170,"alternatives":196},[95,100],{"value":96,"source":52,"evidence":97},"Brings ML into the database, avoiding data export",[98,99],{"id":55,"by":56,"text":57},{"id":66,"by":67,"text":68},{"value":101,"source":52,"evidence":102},"Training from SQL views is a nice approach",[103,107],{"id":104,"by":105,"text":106},"31246299","thejansen","Cool approach. This nicely fits in the trend of SQL-as-much-as-possible because that makes it just a tiny bit more accessible. Definitely going to play with this in the next few…",{"id":108,"by":60,"text":109},"31241087","Yep! Part of the power of being inside the OLTP is that you can just create a VIEW of your training data, which could be anything from customer purchases, search results,…",[111,116,135],{"value":112,"source":52,"evidence":113},"Running ML on the primary OLTP database risks resource contention",[114,115],{"id":55,"by":56,"text":57},{"id":80,"by":81,"text":82},{"value":117,"source":52,"evidence":118},"Logo and name suggest official PostgreSQL affiliation",[119,123,127,131],{"id":120,"by":121,"text":122},"31239704","properdine","You might want to use a different elephant logo... the branding implies that it is an official part of the project. E.g. Postico uses an elephant but not the _same_ elephant -…",{"id":124,"by":125,"text":126},"31240142","macspoofing","You should review the PostgreSQL trademark policy: https:\u002F\u002Fwww.postgresql.org\u002Fabout\u002Fpolicies\u002Ftrademarks\u002F",{"id":128,"by":129,"text":130},"31240664","jmuguy","This logo and name, to me at least, implied some sort of official affiliation with the PostgreSQL project. Which I think should at least be clarified on the site\u002Freadme.",{"id":132,"by":133,"text":134},"31243026","pcj-github","I found this to be very confusing and took some sorting out that this is not an official project of postgresql. I personally think a name change and logo redo is in order here.",{"value":136,"source":52,"evidence":137},"Not usable on managed services like RDS due to extension limits",[138],{"id":139,"by":60,"text":140},"31239795","We can’t control the extensions RDS allows to be installed, and they are historically conservative. Lev and I do have some fairly extensive experience with replication patterns…",[142,149,156,163],{"value":143,"source":52,"evidence":144},"Support for managed services like RDS",[145],{"id":146,"by":147,"text":148},"31239719","LunaSea","Do you plan on adding support for managed PostgreSQL services like RDS in the future?",{"value":150,"source":52,"evidence":151},"Support for older Python versions (3.6)",[152],{"id":153,"by":154,"text":155},"31240483","craigkerstiens","I just gave it a try to install on Crunchy Bridge [1], disclaimer I work at Crunchy Data. I did this as a standard user vs. anything special about access. I got quite close, but…",{"value":157,"source":52,"evidence":158},"Cross-validation and hyperparameter search",[159],{"id":160,"by":161,"text":162},"31240609","kipukun","Neat project. Any roadmap for cross-validation support (GridSearchCV and friends)?",{"value":164,"source":52,"evidence":165},"Custom proprietary model support",[166],{"id":167,"by":168,"text":169},"31242016","obert","is it possible, or how hard is it, to plug in custom proprietary models?",[171,182,189],{"value":172,"source":52,"evidence":173},"How does it compare to MADlib and MindsDB",[174,178],{"id":175,"by":176,"text":177},"31239758","waatels","Hello really nice ! Can you explane the differences with https:\u002F\u002Fmadlib.apache.org\u002F ? Wouldnt an OLAP db better suited than pg for this kind of workload ? Does being a postgreSQL…",{"id":179,"by":180,"text":181},"31244772","dayeye2006","How this compares to https:\u002F\u002Fmindsdb.com\u002F",{"value":183,"source":52,"evidence":184},"How are train\u002Fvalidation\u002Ftest splits and metrics handled",[185],{"id":186,"by":187,"text":188},"31239484","jorgemf","How do you deal with different dataset train\u002Fvalidation\u002Ftest? How do you measure the degradation of the model? Is there any way to select the metric you target (accuracy,…",{"value":190,"source":52,"evidence":191},"Whether GPU training is supported",[192],{"id":193,"by":194,"text":195},"31240546","gabereiser","This is awesome. I’m guessing the models are executed on the database server and not a separate cluster? What about GPU training? How is that handled? I’d love to see more docs.",[197,208,216,220],{"value":198,"source":52,"evidence":199},"BigQuery ML",[200,204],{"id":201,"by":202,"text":203},"31243921","manca","Interesting concept, but I think Big Query ML [1] has been providing similar features for years now. Curious to learn what are the differences, other than offering this as a…",{"id":205,"by":206,"text":207},"31242473","bazhova","This is great! FYI for those who haven't seen, BigQuery can also run statistical learning methods directly on your data as part of the query. Really cool to see ML going this…",{"value":209,"source":52,"evidence":210},"MADlib",[211,212],{"id":175,"by":176,"text":177},{"id":213,"by":214,"text":215},"31243453","dvasdekis","How about the difference between this and the Madlib project? Better ergonomics? I've used Madlib in the past and although it was 'successful', the constraint was unfamiliarity…",{"value":217,"source":52,"evidence":218},"MindsDB",[219],{"id":179,"by":180,"text":181},{"value":221,"source":52,"evidence":222},"Amazon Aurora ML extensions",[223],{"id":224,"by":225,"text":226},"31272734","ttfkam","Amazon Aurora already has ML extensions for connecting to AWS SageMaker and AWS Comprehend. I doubt they'd add another, especially one they didn't write and isn't integrated with…",{"comments":42,"answered":42,"key":228},[],{"commenters":230,"max_depth":46,"maker_replies":42,"first_comment_mins":231,"lifespan_hours":232},46,22,68,[234,236,239,243,245,247,250,252,255,258,259,263,265,268,269,271,273,276,278,280,284,285,286,288,290,292,293,295,296,297,302,305,310,311,314,316,320,322,327,330,333,337,339,342,344,346,349,353,355,360,364,366,368,370,373,377,379,382,386,388,390,393,397,401,404,407,408,411,412,413],[55,42,46,235,56,42],"project\u002Fpositive\u002Fidea",[237,42,46,235,238,42],"31242000","jiocrag",[240,42,42,241,242,42],"31242645","other\u002Fneutral\u002Fnone","coatue",[244,42,42,241,60,42],"31243055",[80,42,33,246,81,42],"project\u002Fcritical\u002Fidea",[248,42,42,249,60,42],"31244633","commenter\u002Fneutral\u002Fnone",[251,42,42,249,81,42],"31244975",[253,42,46,235,254,42],"31245048","chaps",[256,42,42,249,257,42],"31245517","code_biologist",[66,33,46,235,67,42],[260,44,18,261,262,42],"31238811","project\u002Fneutral\u002Fidea","phenkdo",[264,44,18,261,60,42],"31238963",[266,44,18,261,267,42],"31240876","streetcat1",[108,44,46,235,60,42],[146,18,18,270,147,42],"project\u002Fneutral\u002Fnone",[139,18,18,272,60,42],"maker\u002Fneutral\u002Fnone",[274,18,18,261,275,42],"31240145","michelpp",[153,18,18,277,154,42],"project\u002Fneutral\u002Fexecution",[279,18,18,272,60,42],"31240550",[281,18,33,282,283,42],"31246555","project\u002Fcritical\u002Fexecution","prometheon1",[224,18,33,246,225,42],[186,39,18,270,187,42],[287,39,18,272,60,42],"31239956",[289,39,18,272,187,42],"31241742",[291,39,18,272,60,42],"31243535",[205,46,46,235,206,42],[175,294,18,261,176,42],6,[59,294,18,272,60,42],[213,294,18,270,214,42],[298,299,18,300,301,42],"31240661",7,"project\u002Fneutral\u002Fperformance","bguberfain",[303,299,18,304,60,42],"31240742","maker\u002Fneutral\u002Fperformance",[306,307,18,308,309,42],"31273341",8,"project\u002Fmixed\u002Fdocs","Abishek_Muthian",[84,38,18,300,85,42],[312,38,42,313,60,42],"31242335","commenter\u002Fneutral\u002Fperformance",[315,38,42,313,85,42],"31245239",[317,34,33,318,319,42],"31238862","project\u002Fcritical\u002Fdesign","simonw",[201,321,18,261,202,42],11,[323,324,18,325,326,42],"31239051",12,"project\u002Fneutral\u002Fdocs","sagaro",[328,324,46,329,326,42],"31239096","project\u002Fpositive\u002Fdocs",[331,324,18,332,60,42],"31239134","maker\u002Fneutral\u002Fdocs",[334,335,46,235,336,42],"31243074",13,"chartpath",[104,338,46,235,105,42],14,[340,31,33,246,341,42],"31246160","lysecret",[160,343,18,270,161,42],16,[345,343,18,272,60,42],"31240684",[347,343,46,348,60,42],"31268149","project\u002Fpositive\u002Fexecution",[350,351,42,241,352,42],"31244054",17,"tomerbd",[193,354,46,329,194,42],18,[356,357,46,358,359,42],"31243590",19,"project\u002Fpositive\u002Fnone","teknopurge",[361,362,18,300,363,42],"31243406",20,"xpe",[365,362,18,300,60,42],"31243631",[367,362,42,249,363,42],"31247334",[179,369,18,261,180,42],21,[371,231,18,261,372,42],"31246653","wodenokoto",[374,375,18,300,376,42],"31239974",23,"ekzy",[378,375,46,235,60,42],"31240130",[380,375,33,246,381,42],"31241525","noogle",[383,384,46,358,385,42],"31239623",24,"debarshri",[387,384,18,272,60,42],"31239726",[167,389,18,261,168,42],25,[391,389,18,392,60,42],"31242237","maker\u002Fneutral\u002Fidea",[394,395,18,325,396,42],"31243782",26,"toddm",[398,399,42,241,400,42],"31244894",27,"matthewtovbin",[402,403,18,270,129,42],"31238661",28,[405,403,42,249,406,42],"31238825","jpgvm",[120,403,33,318,121,42],[409,403,46,410,376,42],"31240127","project\u002Fpositive\u002Fdesign",[124,403,33,318,125,42],[128,403,33,318,129,42],[132,403,33,318,133,42],[415,416],{"id":4,"t":12,"points":14,"comments":15},{"id":417,"t":418,"points":44,"comments":42},"38294209",1700163054,[420,425,429,434,438,443],{"id":421,"name":422,"points":423,"category":424},"31210680","Radiopaper",346,"consumer_social",{"id":426,"name":427,"points":428,"category":424},"31238958","Hangtime",311,{"id":430,"name":431,"points":432,"category":433},"31192956","Fintopea",245,"finance_crypto",{"id":435,"name":436,"points":437,"category":433},"31247312","betagainst",217,{"id":439,"name":440,"points":441,"category":442},"31225494","NoiseCraft",214,"creative",{"id":444,"name":445,"points":446,"category":447},"31203024","Joyride",211,"dev_tools",false,"v2",{"id":451,"name":5,"line":6,"ai":448,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":452,"top":18,"kw":453,"maker":11,"domain":10,"tier":19,"attempts":44,"strip":454,"url":9,"mix":455,"praise":96,"objection":112,"sub":21},31238396,0.11,[24,25,26,27],"5510553533313353333013551355333515333131",[31,32,33,34],1794036619828]